WO2005000534A1 - 脚式移動ロボットの制御装置 - Google Patents
脚式移動ロボットの制御装置 Download PDFInfo
- Publication number
- WO2005000534A1 WO2005000534A1 PCT/JP2004/009472 JP2004009472W WO2005000534A1 WO 2005000534 A1 WO2005000534 A1 WO 2005000534A1 JP 2004009472 W JP2004009472 W JP 2004009472W WO 2005000534 A1 WO2005000534 A1 WO 2005000534A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- target
- floor reaction
- reaction force
- instantaneous value
- amount
- Prior art date
Links
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B62—LAND VEHICLES FOR TRAVELLING OTHERWISE THAN ON RAILS
- B62D—MOTOR VEHICLES; TRAILERS
- B62D57/00—Vehicles characterised by having other propulsion or other ground- engaging means than wheels or endless track, alone or in addition to wheels or endless track
- B62D57/02—Vehicles characterised by having other propulsion or other ground- engaging means than wheels or endless track, alone or in addition to wheels or endless track with ground-engaging propulsion means, e.g. walking members
- B62D57/032—Vehicles characterised by having other propulsion or other ground- engaging means than wheels or endless track, alone or in addition to wheels or endless track with ground-engaging propulsion means, e.g. walking members with alternately or sequentially lifted supporting base and legs; with alternately or sequentially lifted feet or skid
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J5/00—Manipulators mounted on wheels or on carriages
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J13/00—Controls for manipulators
Definitions
- the present invention relates to a control device suitable for not only walking but also running of a legged moving port.
- a gait target gait
- a gait for the movement of a legged mobile robot, for example, a bipedal robot
- the main purpose has been to generate gaits.
- it has been desired to generate a gait that allows the lopot to perform not only walking but also running.
- it is desired to create a gait that can move the mouth pot smoothly on slippery floors (so-called low mu road) where sufficient frictional force cannot be generated.
- kanji “gait” includes “step”, so it is easy to be misunderstood that it is limited to walking.
- “gait” is originally a word that refers to the running form of a horse, such as a trot. As used, the concept includes running.
- Figure 51 shows the pattern of the body vertical position and the floor reaction force vertical component (sum of the floor reaction force vertical components acting on the left and right legs) in a typical run
- Figure 52 shows the typical The vertical body position and the floor reaction force vertical component pattern during walking are shown.
- the body vertical position speed means the vertical position of the body representative point and its speed.
- the body horizontal position speed is the body The horizontal position of the representative point and its speed.
- the body vertical position speed and the body horizontal position speed are combined and called the body position speed.
- the floor reaction force vertical component is the floor To distinguish it from the moment component around the vertical axis of the reaction force, it should be described as “translational floor reaction force vertical component", but since the word becomes longer, "translation” is omitted here.
- Translational floor reaction force horizontal component is also described as “floor reaction force horizontal component” with “translation” omitted.
- the upper body is the highest at the moment when the upper body passes over the supporting legs, and the lowest in running at this moment, that is, in walking and running, The phase of the upper / lower movement pattern is reversed.
- the floor reaction force is relatively constant in walking, while it fluctuates greatly in running, and the floor reaction force is greatest at the moment when the upper body passes over the supporting legs. And, naturally, the floor reaction force is zero at the moment when all the legs are in the air at the same time.
- a closer observation shows that a running reaction generates a floor reaction force that is approximately proportional to the amount of contraction of the support leg. In other words, in running, it can be said that the legs are jumping using the legs as springs.
- Slow jogging has the same upper and lower body movement phase as a typical run. Also, in slow jogging, there is no moment when all the legs are in the air at the same time. In many cases, the floor reaction force is not completely zero but almost zero at the moment when the support leg and the free leg are switched.
- running is a movement mode in which the floor reaction force becomes 0 or almost 0 at the moment when the support leg is switched. It can be said that it is a moving form (the floor reaction force vertical component is relatively constant).
- the applicant of the present application has proposed the following technology in Japanese Patent Application Laid-Open No. Hei 5-337849.
- the actual ZMP is shifted.
- the motion of the desired gait to generate a moment in the posture recovery direction within the allowable range around the target ZMP and to generate a moment around the target ZMP on the model used to generate the desired gait To determine.
- intentionally generating a moment on the model it can be seen on the actual robot. It is possible to obtain the same effect as generating the momentum in the recovery direction.
- the floor reaction force vertical component sometimes becomes 0 or close to 0, so at that time, even if the friction coefficient is high, the limit of the moment vertical component of the floor friction force is limited. Becomes or approaches zero. Therefore, the floor reaction force moment vertical component of the desired gait exceeded the limit, and there was a risk of spinning and falling.
- the present applicant first shakes the arm so as to cancel the moment vertical component that occurs in the target gait other than the arm. Suggested what was done. In this case, the moment vertical component of the desired gait is almost 0, but when the leg is moved vigorously, the arm swings violently. Generally, in a humanoid mouth pot, the mass of the arm is smaller than the mass of the leg. Therefore, to completely cancel the vertical component of the moment, the arm must swing more vigorously than the swing of the legs.
- the center of the arm swing gradually offsets, and the swing of the left and right arms becomes asymmetric.
- the center of the arm swing gradually offsets, and the swing of the left and right arms becomes asymmetric.
- the angular momentum change due to the leg and torso turning to the left In order to counteract the left arm, the left arm swings large toward the front and swings small toward the back, and the right arm swings toward the front and small behind. Therefore, in this case, the left arm may reach the front swing motion limit, and the right arm may reach the rear swing motion limit.
- an object of the present invention is to solve the above-mentioned problems and to provide a control device capable of generating a more suitable gait irrespective of a gait form such as walking or running or a friction state of a floor surface. Is to provide.
- the present invention considers the limit of the moment vertical component of the frictional force between the robot and the floor surface, and stabilizes the posture state of the robot in one direction.
- An object of the present invention is to provide a control device that can operate a robot with a gait that can prevent spin and falling of the robot due to the spin while keeping the spin. It is another object of the present invention to provide a control device that can operate a robot with a gait motion pattern that satisfies dynamic equilibrium conditions even in the air or when the limit of the moment vertical component of frictional force is extremely low. And Furthermore, the object is to prevent the left-right asymmetry of the desired gait from becoming large, and to ensure the continuity of the motion. Disclosure of the invention
- the first invention of the gait generating device for a legged mobile robot of the present invention is:
- the instantaneous value of the target motion and the target floor reaction force of the leg-type moving robot that moves by moving the leg extended from the upper body is used to express at least the relationship between the motion of the robot and the floor reaction force.
- a legged mobile robot control device for controlling the operation of the robot so as to follow the determined target motion and the instantaneous value of the target floor reaction force while sequentially determining the target motion and the determined floor reaction force using the dynamic model.
- the vertical component of the floor reaction force moment or the component of the floor reaction force moment in the direction normal to the floor surface to be acted on by the robot operating following the target motion and the target floor reaction force is defined as a limiting target amount.
- At least a limit corresponding to the target floor reaction force based on a deviation between a target state quantity and an actual state quantity of the lopot with respect to a posture of the mouth pot about a vertical axis or a floor normal axis, and the allowable range. While keeping the target amount within the allowable range, the difference between the floor reaction chamoment balanced with the target motion on the dynamic model and the floor reaction force moment of the target floor reaction force approaches the deviation to zero. And a target instantaneous value determining means for determining an instantaneous value of a desired motion and a desired floor reaction force.
- the target floor is determined based on at least the deviation between the target state quantity related to the posture of the robot about the vertical axis or the floor normal axis and the actual state quantity of the robot and the allowable range.
- the difference between the floor reaction chamomile that balances the target motion with the dynamic model on the dynamic model and the floor reaction chamomile of the target floor reaction force while keeping the restriction target amount corresponding to the reaction force within the allowable range. Determines the instantaneous values of the target motion and the desired floor reaction force so that the deviation approaches zero.
- the state quantity relating to the posture of the mouth pot about the vertical axis or the floor normal axis (hereinafter referred to as the posture state quantity in one direction) is referred to as the target state quantity-that is, the target posture state quantity in the Y direction in the target motion.
- the target state quantity-that is, the target posture state quantity in the Y direction in the target motion While keeping the deviation close to 0 (while keeping the deviation close to 0), it is possible to limit the restricted target amount such as the vertical component of the floor reaction force moment acting on the actual mouth pot within the allowable range. That is, the instantaneous values of the target motion and the target floor reaction force are determined in consideration of the limit of the vertical component of the frictional moment and the deviation.
- the target instantaneous value determination means includes: a means for determining a compensation floor reaction moment, which is an additional floor reaction chamois, for bringing the deviation closer to 0 in accordance with the deviation;
- the predetermined temporary instantaneous value of the target motion is adjusted in accordance with the floor reaction chamoment and the compensated floor reaction force moment that are balanced on the dynamic model so that the limited target amount does not exceed the allowable range.
- Means for determining a correction amount of the instantaneous value It is preferable to determine the instantaneous value of the target motion by correcting the temporary instantaneous value according to the positive amount (second invention).
- a floor reaction force moment (balanced with the target motion) generated on the dynamic model by the predetermined tentative instantaneous value of the target motion, and a compensation floor reaction camo for making the deviation close to zero. Since the correction amount of the predetermined provisional instantaneous value is determined in accordance with one comment, the amount of restriction target is set within an allowable range while the amount of posture state of the robot in one direction is close to the target state amount. The appropriate instantaneous value of the target motion can be determined.
- the second invention includes means for determining a model corrected floor reaction force moment that is an additional floor reaction force moment for bringing the state quantity of the dynamic model closer to a predetermined state quantity,
- the means for determining the correction amount of the predetermined temporary instantaneous value of the target motion includes at least the floor reaction force moment, the compensation floor reaction force moment, and the model correction that are balanced on the dynamic model with the predetermined temporary instantaneous value. It is preferable to determine the correction amount of the provisional instantaneous value of the target movement so that the restriction target amount does not exceed the allowable range in accordance with floor anti-moment (third invention).
- the dynamic model when determining the correction amount of the predetermined temporary instantaneous value of the target motion, the floor reaction force moment balanced on the dynamic model with the predetermined temporary instantaneous value of the target motion, and the compensation floor reaction Since not only the force moment but also the model corrected floor reaction force moment relating to the state quantity of the dynamic model is considered, the dynamic model can be prevented from diverging in addition to the effect of the second invention.
- the correction amount of the predetermined temporary instantaneous value is a correction amount of a motion that changes a vertical component or a floor surface normal direction component of the angular momentum change rate of the lopot. Is preferable (the fourth invention).
- the motion for changing the vertical component or the floor surface normal direction component of the angular momentum change rate of the robot is the upper body of the robot and / or the arm body extended from the upper body. It is preferable that the exercise is performed (fifth invention). 'According to this, it is easy to correct the temporary instantaneous value of the target motion.
- the motion that changes the vertical component or the floor normal direction component of the rate of change in angular momentum of the robot is a motion that maintains a constant center of gravity of the robot.
- the instantaneous value of the target motion of the legged locomotion locomotion is calculated using a dynamic model that expresses at least the relationship between the locomotion and the floor reaction force.
- the control device of the legged mobile robot which controls the operation of the robot so as to follow the determined instantaneous value of the target motion while sequentially determining
- the vertical component of the floor reaction force moment or the component of the floor reaction force moment in the direction normal to the floor surface to be acted on the lopot operating following the target motion is defined as the limited target amount, and the allowable range of the limited target amount is defined.
- the instantaneous value of the target motion is determined so that the limit target amount determined in accordance with the floor reaction chamoment balanced with the target motion on the dynamic model and the compensated floor reaction force moment falls within the allowable range. And a target instantaneous value determining means.
- fishing is performed on the target motion on the dynamic model.
- the above-mentioned limited amount (e.g., the floor reaction chamoment and the compensation floor reaction balanced with the target motion on the dynamics model) determined according to the floor reaction force moment that fits and the compensation floor reaction camoment.
- the instantaneous value of the target motion is determined such that the sum of the moment and the force moment falls within the allowable range. For this reason, even if a compensating floor anti-chamoment is additionally generated to bring the posture state quantity of the mouth pot in the yaw direction closer to the target state quantity (posture state state quantity in the target movement). It is possible to generate a target motion that can limit the amount of limiting such as the vertical component of the floor reaction force moment acting on the robot within the allowable range.
- the instantaneous value of the target motion is determined in consideration of the limit of the vertical component of the frictional moment and the deviation.
- the target movement of the robot which can prevent the spin of the mouth pot and the overturn caused by the rotation while keeping the posture state amount of the robot in the target direction of the target movement stably, can be achieved.
- the amount to be restricted does not always need to be set to 0 or almost 0. Can be prevented.
- a means for controlling the operation of the robot so that the restricted object amount within the allowable range is set as a target floor reaction force moment and the target floor reaction force moment is followed. 7th invention). According to this, it is possible to prevent the spin of the lopot and the overturn caused by the spin while keeping the posture amount of the lopot in one direction stably at the posture state amount of the target motion. In addition, it is possible to prevent the movement trajectory of the mouth pot from deviating from the trajectory of the target gait, and to prevent the continuous stability of the posture of the mouth pot in one direction from being impaired.
- the target instantaneous value determination means may include the limited target amount determined in accordance with a floor reaction chamoment balanced with the target operation on the dynamic model and the compensation floor reaction force moment.
- the motion that changes the vertical component or the floor normal direction component of the rate of change in the angular momentum of the mouth pot among the motions of the mouth pot is adjusted. It is preferable to determine the instantaneous value of (the eighth invention).
- the instantaneous value of the target motion can be determined so that the amount to be restricted falls within the allowable range.
- the motion for changing the vertical component or the floor normal direction component of the angular momentum change rate of the lopot is based on the upper body and the nose of the lopot or the arm extending from the upper body.
- the exercise is preferably (ninth invention).
- the motion for changing the vertical component or the floor normal direction component of the angular momentum change rate of the robot is a motion that keeps the entire S center of the robot constant.
- a tenth invention of the control device for a legged mobile robot includes an instantaneous value of a target motion of the legged mobile robot which moves by moving a leg extended from the upper body, at least.
- a legged movement for controlling the operation of the robot so as to follow the determined instantaneous value of the target movement while sequentially determining using a dynamic model representing the relationship between the movement of the robot and the floor reaction force.
- the vertical component of the floor reaction force moment to be applied to the robot operating following the target motion or the component of the floor reaction force moment in the direction normal to the floor surface is set as the limit target amount, and the limit target amount is allowed.
- a tolerance setting method for setting the range
- Tentative instantaneous value determining means for sequentially determining a tentative instantaneous value of the target motion; At least an additional floor camber for approaching the deviation to zero in accordance with the deviation between the target state quantity and the actual state quantity of the robot with respect to the posture of the robot about the vertical axis or the floor normal axis.
- Means for determining a compensation floor reaction force moment which is a compensation floor reaction force moment,
- a moment correction is made for the amount of deviation from the permissible range of the limited target amount determined according to the floor reaction force moment and the compensated floor reaction force moment that are balanced on the dynamic model with the provisional instantaneous value of the target motion.
- a target instantaneous value determination that determines the instantaneous value of the target motion by correcting the provisional instantaneous value of the target motion so that the deviation tends to decrease in accordance with the momentum. Means.
- the limited target amount (for example, the target value) determined in accordance with the floor reaction force moment and the compensating floor reaction moment that balance the provisional instantaneous value of the target motion on the dynamic model with the dynamic model.
- the sum of the floor reaction force moment 'balanced on the dynamic model with the provisional instantaneous value of the motion and the compensation floor reaction force moment) that deviates from the permissible range is defined as the moment correction operation amount.
- the instantaneous value of the target motion is determined by correcting the provisional instantaneous value of the target motion so that the deviation tends to decrease in accordance with the amount of operation for correcting the target.
- the target motion that can restrict the amount of restriction, such as the vertical component of the floor reaction force moment acting on the robot, within an allowable range. That is, the instantaneous value of the target motion is determined in consideration of the limit of the vertical component of the frictional moment and the deviation. As a result, it is possible to prevent the mouth pot from spinning and falling due to the spin of the mouth pot while maintaining the posture state quantity of the mouth pot in the direction of the target movement stably at the posture quantity of the target motion. Can determine the target exercise Wear.
- the target instantaneous value determining means obtains a correction amount of the provisional instantaneous value of the target motion in accordance with the moment correction operation amount passed through a mouth-to-pass fill, and calculates the obtained correction amount. It is preferable to determine the instantaneous value of the target motion by correcting the tentative instantaneous value according to (1st invention).
- the limit target amount determined in accordance with the floor reaction force moment and the compensating floor reaction force moment that are balanced on the dynamic model with the provisional instantaneous value of the target movement is set in the allowable range.
- a means for controlling the robot operation so that the floor reaction force moment corresponding to the limited amount to be restricted, which is limited by the above, is set as the target floor reaction force moment, and the robot follows the target floor reaction camoment. Is preferable (the 12th invention).
- the twelfth aspect it is possible to prevent the spin of the lopot and the overturn caused by the spin while keeping the posture state amount of the mouth pot in the yaw direction at the posture state amount of the target operation stably.
- the target instantaneous value determination means is configured to change a vertical component or a floor normal direction component of the angular momentum change rate of the robot. It is preferable to determine the instantaneous value of the target motion by correcting the motion from the temporary instantaneous value of the target motion (a thirteenth invention).
- the instantaneous value of the target motion can be determined so that the amount to be restricted falls within the allowable range.
- the motion for changing the vertical component or the floor normal direction component of the angular momentum change rate of the mouth pot is extended from the upper body and Z or the upper body of the robot. It is preferable that the exercise be performed for the arms (first to fourteenth inventions).
- the state quantity relating to the posture of the robot is, specifically, an angle of the upper body of the robot (rotation angle in one direction) or an angular velocity of the robot. (The angular velocity of the rotation angle in one direction) (15th invention). According to the fifteenth invention, it is possible to enhance the stability of the posture of the upper body of the mouth pot in one direction. This also applies to the fifteenth to fifty-first inventions described later.
- a slip judging means for judging occurrence of slip of the mouth pot, and the allowable range setting means sets the allowable range according to a judgment result of the slip judging means. It is preferable to set variably (the 16th invention).
- the occurrence of slippage of the robot (a robot operating following the instantaneous value of the target motion or the instantaneous values of the target motion and the target floor reaction force) is determined. Since the allowable range of the restricted amount is variably set according to the result, the occurrence of slippage of the mouth pot can be reliably suppressed. When it is determined that slippage occurs, the allowable range should be set so as to narrow it.
- the occurrence of slip is determined as follows. Can.
- the slip judging means judges the occurrence of slip based on at least the ground speed of the tip of the leg that is in contact with the ground (17th invention). In this case, for example, when the absolute value of the ground speed is larger than a predetermined value, it can be determined that a slip has occurred.
- the slip determination means obtains an apparent spring constant of the leg based on at least a temporal change rate of an actual floor reaction force acting on the grounded leg and a ground speed of the tip of the leg.
- the slip determination means determines the occurrence of slip based on at least the actual floor reaction force acting on the grounded leg passed through a band-pass filter having a frequency passing characteristic in a range near a predetermined frequency.
- the actual floor reaction force passed through the band-pass filter corresponds to the vibration component of the actual floor reaction force when so-called sliding vibration occurs. Then, for example, when the magnitude (absolute value) of the vibration component is larger than a predetermined value, it can be determined that the slip vibration has occurred.
- the determination of the occurrence of slip can be made in any of the seventeenth to nineteenth inventions, but the determination of the occurrence of slip is made by combining two or more of the seventeenth to nineteenth inventions. You may do so.
- a 20th invention of the legged mobile robot control device of the present invention generates a desired gait of a legged mobile robot that moves by moving a plurality of legs extended from the upper body,
- the control device for controlling the movement of the lopot so as to follow the desired gait At least according to the deviation between the target state quantity relating to the posture of the robot around the vertical axis or the floor surface normal axis and the actual state of the robot, an additional floor countermeasure for bringing the deviation closer to 0 Means for determining a compensating floor reaction force moment,
- the vertical component of the floor reaction force moment or the floor component normal direction component of the floor reaction force moment to be applied to the lopot operating following the target gait is set as the target object amount, and the allowable amount of the object amount is limited.
- a tolerance setting method for setting the range
- At least the provisional instantaneous value of the target motion is input to a dynamic model representing the relationship between the motion of the robot and the vertical component or the floor normal direction component of the moment of the floor reaction force balanced with the motion.
- a vertical component or a floor normal direction component of the floor reaction force moment as an output of the dynamic model is obtained, and a predetermined function is obtained from the obtained vertical component or the floor surface normal component and the compensation floor reaction force moment.
- Model calculation means for determining a model limit target amount instantaneous value which is a provisional instantaneous value of the limit target amount by calculation,
- Target instantaneous value determination means for determining the instantaneous value of the target motion by correcting the tentative instantaneous value of the target motion so that at least the model limited target amount instantaneous value falls within the allowable range.
- a vertical component or a floor normal direction component of a floor reaction force moment as an output of the dynamic model is obtained. Is required. Then, a predetermined operation (for example, addition) is performed from the obtained vertical component or floor normal direction component and the compensated floor reaction force moment to obtain a model restriction target amount instantaneous value (this is a provisional instantaneous value of the restriction target amount).
- a model restriction target amount instantaneous value is a provisional instantaneous value of the restriction target amount.
- a compensating floor reaction force moment for generating the posture state quantity of the mouth pot in the gutter direction closer to the target state quantity (posture state quantity in the one direction in the target motion) is additionally generated. It is possible to generate a target motion that can limit the amount of restriction such as the vertical component of the floor reaction force moment acting on the actual robot within an allowable range. That is, the instantaneous value of the target motion is determined in consideration of the limit of the vertical component of the frictional moment and the deviation. This makes it possible to prevent the spin of the robot and the overturn caused by it, while maintaining the posture of the robot in one direction stably at the posture of the target motion. Can be determined.
- the dynamic model is basically one having a high approximation accuracy.
- the target instantaneous value determining means includes a vertical component or a floor normal direction component of a moment of a floor reaction force substantially balanced with the instantaneous value of the target motion on the dynamic model.
- the target instantaneous value determining means includes a perturbation representing a relationship between a perturbation motion of the robot and a vertical component of a floor reaction force moment or a perturbation component of a floor surface normal component.
- a model means for determining a perturbation model operation amount for operating a perturbation component of the perturbation model based on at least the determined model limit target amount instantaneous value and the permissible range; and determining the determined perturbation model operation amount Means for obtaining a correction amount of the target motion by inputting to a perturbation model; and means for determining an instantaneous value of the target motion by correcting a provisional instantaneous value of the target motion by the correction amount.
- Preferred (22nd invention) means for determining an instantaneous value of the target motion by correcting a provisional instantaneous value of the target motion by the correction amount.
- the perturbation model since the perturbation model has a high linearity, it is easy to calculate an appropriate correction amount of the target movement for keeping the limited target amount within the allowable range, and to temporarily calculate the target movement based on the correction amount.
- the instantaneous value correction process becomes easy.
- the means for determining the perturbation model manipulated variable includes at least the estimated value of the restricted subject quantity when the perturbation model manipulated variable is assumed to be 0, and at least the determined model restricted subject quantity instantaneous value. And a means for comparing the determined estimated value with the allowable range and determining a limited amount to be restricted within the allowable range based on the comparison. It is preferable to determine the perturbation model operation amount based on the difference between the determined model restriction target amount instantaneous value and the restricted restriction target amount. invention).
- an estimated value of the restricted target amount is obtained when the perturbation model operation amount is assumed to be 0, and the restricted restricted target amount is determined based on a comparison between the estimated value and the allowable range.
- the model restricted target amount instantaneous value may be directly used as the estimated value of the restricted target amount, but the estimated value of the restricted target amount may be determined in consideration of the gyro effect.
- the restricted amount to be restricted is determined as close to the estimated value of the restricted amount as possible within the allowable range.
- a perturbation model operation amount is determined based on at least the difference between the instantaneous value of the model restriction target amount and the restricted restriction target amount.
- the means for determining the perturbation model operation amount ⁇ in accordance with at least a state amount of the perturbation model, and the means for determining the perturbation model operation amount is at least the determination. It is preferable to determine a perturbation model operation amount to be input to the perturbation model based on the model restriction target amount instantaneous value, the permissible range, and the required value (the 24th invention).
- the means for determining the required value of the perturbation model manipulated variable includes: sequentially determining the required value according to a feedback control law according to a deviation between a state quantity of the perturbation model and a target value for the state quantity. It is preferable to Yes (25th invention).
- the perturbation model operation amount is determined such that the state amount of the perturbation model is substantially maintained near a certain target state amount. This makes it possible to make the correction amount of the target motion determined by each perturbation model suitable for keeping the amount to be restricted within an allowable range while keeping the state of the perturbation model stable.
- the means for determining the perturbation model operation amount includes an estimated value of the restriction target amount when it is assumed that the perturbation model operation amount matches the required value.
- Means for determining the restricted restriction target amount and preferably determines the perturbation model manipulated variable based on at least the difference between the model restriction target amount instantaneous value determined above and the restricted restriction target amount. 26 inventions).
- the perturbation model operation amount is made to match the required value, that is, an estimated value of the restricted object amount is obtained when only the required value is input to the perturbation model.
- Determine the restricted quantity determined based on the comparison between the value and the allowable range.
- the sum of the instantaneous value of the model restriction target amount and the required value may be used as the estimation value of the restriction target amount as it is, but the estimated value of the restriction target amount is determined in consideration of the gyro effect. You may do so.
- the restricted amount to be restricted is determined to be a value as close as possible to the estimated value of the restricted amount within the allowable range.
- a perturbation model operation amount is determined based on at least a difference between the model restriction target amount instantaneous value and the restricted restriction target amount.
- the target instantaneous value determination means may include means for additionally inputting the correction amount of the target movement to the dynamic model (a twenty-seventh invention).
- the vertical component or the floor normal direction component of the floor anti-chamoment balanced with the instantaneous value from the dynamic model to the target motion can be obtained directly.
- the perturbation model includes a perturbation motion for perturbing a vertical component or a floor normal direction component of a rate of change in angular momentum of the robot and the restriction object. It is preferable that the model represents the relationship with the quantity perturbation (28th invention).
- the perturbation is a perturbation that keeps the position of the center of gravity of the lopot substantially constant (the twenty-ninth invention).
- the target motion can be corrected by the perturbation model without affecting the translational floor reaction force of the mouth pot.
- the perturbation movement is a 'perturbation movement of the upper body and Z of the lopot or an arm extending from the upper body (30th invention).
- the third invention of the legged mobile robot control device of the present invention generates a desired gait of a legged mobile robot that moves by moving a plurality of legs extended from the upper body,
- a control device for controlling the operation of the lopot so as to follow the desired gait At least according to the deviation between the target state quantity related to the posture of the lopot around the vertical axis or the floor surface normal axis and the actual state of the lopot, additional floor anti-chambering to make the deviation close to 0 Means for determining a compensating floor reaction moment,
- Target floor reaction force temporary instantaneous value determining means for sequentially determining at least a temporary instantaneous value of the desired floor reaction force of the target exercise and the desired floor reaction force constituting the desired gait;
- First model calculation means for inputting the values to a first dynamic model representing the relationship between the lopot motion and the floor reaction force, and sequentially obtaining temporary momentary values of the target motion as an output of the first dynamic model;
- the second dynamic model representing the relationship between the motion of the robot and the vertical component or the floor surface normal component of the moment of the floor reaction force balanced with the motion includes at least the provisional instantaneous value of the target motion.
- Second model calculating means for determining a model restricted amount instantaneous value is a provisional instantaneous value of the quantity, '
- a floor reaction chamoment correction amount of a target floor reaction force is determined so that at least the model limited target amount instantaneous value falls within the allowable range, and the determined floor reaction force moment correction amount is determined by the first dynamic model.
- a first model input correction means for additionally inputting to the
- the instantaneous value of the target motion is determined based on at least the input of the second dynamic model.
- the vertical motion of the floor reaction camo as an output of the second dynamics model is obtained.
- a component or floor surface normal direction component is obtained, and a model limiting target amount instantaneous value (temporary instantaneous value of the limiting target amount) is calculated by a predetermined operation (for example, addition) from the compensating floor reaction force moment. This is equivalent to the instantaneous value of the limited amount when the compensating floor reaction force moment is added to the floor reaction force moment at which the desired motion occurs on the dynamic model).
- a floor reaction chamoment correction amount of the target floor reaction force is determined so that at least the model limiting target amount instantaneous value falls within the allowable range, and the determined floor reaction force moment correction amount is determined by the first dynamics. Since the input is additionally provided to the model, based on at least the input of the second kinetic model, an instantaneous value of the target motion that can keep the limited object amount within an allowable range is obtained (for this reason, Even if the compensating floor reaction force moment for additionally bringing the posture state quantity of the pot in one direction closer to the target state quantity (the posture state quantity in the first direction in the target motion) is generated, the actual It is possible to generate a target motion that can limit the amount of restriction such as the vertical component of the floor reaction force moment acting on the pot within the allowable range, that is, by considering the limit of the vertical component of the friction force moment and the deviation.
- the instantaneous value of the target movement is determined, so that the posture of the lopot in one direction is stably maintained at the posture of the target movement, and the spin of the mouth pot and the resulting fall are prevented.
- the robot's movement trajectory can be deviated from the desired gait trajectory, and the robot's continuous stability in one direction can be determined. It is possible to determine the target operation that can prevent damage, and since it is not necessary to always set the restricted amount to 0 or almost 0 as long as the restricted amount falls within the allowable range, the robot This can prevent excessively intense exercise in areas with The
- the second dynamic model is basically one having a higher approximation accuracy than the first dynamic model.
- the vertical component or floor normal direction component of the moment of the floor reaction force substantially balanced with the instantaneous value of the target motion on the second dynamics model, and the compensation floor reaction force moment Means for determining a floor reaction force moment instantaneous value corresponding to the limited target amount instantaneous value determined by the predetermined calculation as the floor reaction force moment instantaneous value of the desired floor reaction force constituting the target gait.
- the first model input correction means estimates the limit target amount of the basal assuming that the floor reaction force moment correction amount of the target floor reaction force is at least zero.
- Means for determining a target amount it is preferable to determine the floor reaction force moment correction amount based on at least the difference between the determined model limit target amount instantaneous value and the limited limit target amount. (33rd invention).
- an estimated value of the limited target amount at least when the floor reaction force moment correction amount of the target floor reaction force is assumed to be 0 is determined, and the estimated value and the allowable range are determined.
- the restricted amount to be restricted is determined based on the comparison with.
- the estimated value may be used, the estimated value of the limited target amount may be determined in consideration of the Jai mouth effect.
- the limited amount to be restricted is determined to be as close as possible to the estimated value of the amount to be restricted within the permissible range. Based on the difference between the two, the amount of floor reaction force moment correction is determined. Even if a compensating floor reaction force moment is additionally generated, it is possible to determine an appropriate floor reaction chamoment correction amount that enables the restriction target amount to be kept within the allowable range.
- the first model input correction means comprises Both the target floor reaction force and the floor reaction duck
- the floor reaction force moment correction amount is determined based on the difference from the limited amount to be restricted (the 34th invention).
- the floor reaction chamoment correction amount of the target floor reaction force is made to match the required value, that is, only the required value is added to the first dynamics model.
- An estimated value of the restricted target quantity when it is assumed to be input is obtained, and the restricted restricted target quantity determined based on a comparison between the estimated value and the allowable range is determined.
- the sum of the instantaneous value of the model restriction target quantity and the above-mentioned required value may be used as the estimation value of the restriction target quantity as it is.
- the restricted amount to be restricted may be as small as possible within the allowable range. It is preferable to determine a value close to the estimated value.
- the floor reaction force moment correction amount is determined based on at least the difference between the model restricted target amount instantaneous value and the restricted restricted target amount.
- the appropriate amount of floor reaction chamoment correction is determined so that the amount to be restricted does not exceed the allowable range and approaches the required value as much as possible. Can be determined.
- a correction amount of the target motion is obtained at least based on the determined model limit target amount instantaneous value and the allowable range, and the obtained correction amount is used as the second power.
- a second model input correction means for additionally inputting to the scientific model may be provided (the 35th invention).
- a perturbation model representing the relationship between the perturbation of the lopot and the perturbation of the restricted object amount, and a floor reaction force moment based on at least the determined model limited object amount instantaneous value and the permissible range.
- the operation amount of the floor reaction force moment is distributed to the floor reaction force moment correction amount input to the first dynamic model and the perturbation model operation amount input to the perturbation model.
- Inputs suitable for the characteristics of the model and the perturbation model can be input to each model. That was Therefore, it is possible to appropriately determine the instantaneous value of the target motion that can keep the limited object amount within an allowable range while preventing the first dynamics model and the perturbation model from becoming unstable.
- the perturbation model it becomes easy to calculate the correction amount of the target motion that is additionally input to the second dynamic model.
- the floor reaction force moment correction amount input to the first dynamic model is desirably the low-frequency component (DC component) of the floor reaction force moment operation amount, and the perturbation model operation amount input to the perturbation model. Is preferably the high frequency component of the manipulated variable of the floor reaction force moment.
- means for determining a required value of the manipulated variable of the floor reaction force moment according to at least the state quantity of the perturbation model is provided, and the means for determining the manipulated variable of the floor reaction force moment is Preferably, an operation amount of a floor reaction force moment to be given to the distribution means is determined based on at least the determined model limit target amount instantaneous value, the allowable range, and the required value (third invention). '
- the thirty-seventh aspect not only the model limiting target amount instantaneous value and the allowable range in consideration of the compensation floor reaction force moment, but also the state quantity of the perturbation model (the rotation angle of the rotating body as an element of the perturbation model, In consideration of the angular velocity, the manipulated variable of the floor reaction force moment input to the distribution means is determined, so that the state quantity of the perturbation model does not depart from the state quantity corresponding to the required value. It is possible to prevent the correction amount of the target motion obtained by the perturbation model from becoming inappropriate.
- the means for determining the required value of the manipulated variable of the floor reaction force moment is based on a feedback control law according to a deviation between a state quantity of the perturbation model and a target value for the state quantity. It is preferable to determine the request value sequentially (38th invention).
- the state quantity of the perturbation model is close to a certain target state quantity.
- the manipulated variable of the floor anti-camo, and consequently, the manipulated variable of the perturbation model is determined so that it is almost maintained by the side. This makes it possible to make the correction amount of the target motion determined by the perturbation model suitable for keeping the state of the perturbation model stable and for keeping the restriction amount within an allowable range.
- the perturbation model includes a perturbation motion for perturbing a component of the rate of change of the angular momentum of the robot around a vertical axis or a floor normal axis.
- the model is preferably a model representing the relationship with the perturbation of the restricted object (the 39th invention).
- the perturbation movement is a perturbation movement for maintaining the position of the center of gravity of the mouth pot substantially constant (the 40th invention).
- the target motion can be corrected by the perturbation model without affecting the translational floor reaction force of the mouth pot.
- the perturbation movement is a perturbation movement of the upper body and Z of the robot or an arm extending from the upper body (the 41st invention). .
- the means for determining the manipulated variable of the floor reaction force moment comprises: Means for determining based on the determined instantaneous value of the limited amount of the model, comparing the determined estimated value with the allowable range, and determining the limited limited amount limited to the allowable range based on the comparison. Means for operating the floor reaction force moment based on at least the difference between the determined instantaneous value of the model restriction target amount and the restricted restriction target amount (the 42nd invention). PC leak 004/009472
- an estimated value of the restricted object amount is obtained when the perturbation model operation amount is assumed to be 0, and the restricted restriction is determined based on a comparison between the estimated value and the allowable range.
- Determine the target amount In this case, for example, the instantaneous value of the model restriction amount may be used as the estimated value of the restriction target amount as it is, but the estimated value of the restriction target amount may be determined in consideration of the gyro effect.
- the restricted amount to be restricted is determined as close to the estimated value of the restricted amount as possible within the allowable range.
- an operation amount of the floor reaction force moment is determined based on at least a difference between the model restriction target amount instantaneous value and the restricted restriction target amount. As a result, even if a compensating floor reaction force moment is additionally generated, the operation amount of the floor reaction chamoment can be determined so that the restriction target amount does not exceed the allowable range.
- the means for determining the operation amount of the floor reaction force moment includes: Means for determining an estimated value of the limited object amount based on at least the determined model limited object amount instantaneous value and the required value, assuming that the estimated value is matched with the required value; and Means for determining a limited amount to be restricted, which is limited to the allowable range based on the comparison, wherein at least the determined instantaneous value of the limited amount of model and the limited amount of restricted It is preferable to determine the operation amount of the floor reaction force moment based on the difference ”(the 43rd invention).
- the perturbation model operation amount is made to match the required value, that is, an estimated value of the restricted object amount is obtained when only the required value is input to the perturbation model.
- Determine the restricted quantity determined based on the comparison between the value and the allowable range.
- the sum of the instantaneous value of the model restriction target amount and the above required value may be used as the estimated value of the restriction target amount as it is, but the estimated value of the restriction target amount is determined in consideration of the gyro effect. May be determined.
- the restricted amount to be restricted is determined to be a value as close as possible to the estimated value of the restricted amount within the allowable range.
- the operation amount of the floor reaction force moment is determined based on at least the difference between the model restricted target amount instantaneous value and the restricted restricted target amount, that is, the portion of the model restricted target amount instantaneous value that deviates from the allowable range. .
- the operation amount of the floor reaction force moment which does not exceed the allowable range and ensures the stability of the perturbation model, is appropriate. Can be determined.
- a fourth invention of the control device for a legged mobile robot according to the present invention generates a desired gait of a legged mobile robot that moves by moving a plurality of legs extended from the upper body,
- a control device for controlling the operation of the robot so as to follow the desired gait In a control device for controlling the operation of the robot so as to follow the desired gait,
- the vertical component of the floor reaction force moment or the floor component normal direction component of the floor reaction force moment to be applied to the lopot operating following the target gait is set as the target object amount, and the allowable amount of the object amount is limited.
- a tolerance setting method for setting the range
- Target floor reaction force temporary instantaneous value determining means for sequentially determining at least a temporary instantaneous value of the desired floor reaction force of the target motion and the desired floor reaction force constituting the desired gait; Values into a first dynamic model representing the relationship between the movement of the lopot and the floor reaction force, and a first model calculation for obtaining a first temporary instantaneous value of the target motion as an output of the first dynamic model Means for determining a second temporary instantaneous value of the target motion, JP2004 / 009472
- the second restricted instantaneous value of the target motion is determined based on at least the temporary instantaneous value of the target floor reaction force so that the instantaneous value of the limited target amount obtained by the predetermined calculation from the above is within the allowable range.
- Model calculation means
- Operation amount calculating means for obtaining an operation amount of a floor reaction force moment based on at least a difference between the first provisional instantaneous value and the second provisional instantaneous value of the target motion so that the difference approaches zero.
- Model input correction means for additionally inputting the operation amount of the floor reaction force moment to at least one of the first dynamic model and the second dynamic model
- a second temporary instantaneous value of the target exercise is determined as a target instantaneous value of the target exercise.
- the tentative instantaneous value of the desired floor reaction force is input to the first dynamic model, and the first tentative instantaneous value of the target motion is obtained. It is also input to the second dynamic model, and the second temporary instantaneous value of the target motion is obtained using the second dynamic model.
- a predetermined component is obtained from the vertical component or the floor normal direction component of the floor reaction force moment balanced with the second tentative instantaneous value of the target operation on the second dynamic model and the compensated floor reaction force moment.
- the second provisional instantaneous value of the target motion is determined so that the restricted target amount instantaneous value obtained by the calculation falls within the allowable range.
- the manipulated variable of the floor reaction force moment is determined so that the difference approaches zero. It is additionally input to at least one of the model and the second kinetic model. Further, the second temporary instantaneous value of the target exercise is different from the target instantaneous value of the target exercise. Is determined.
- the second provisional instantaneous value of the target motion obtained by the restricted second model calculation means has high stability while keeping the restricted target amount within the allowable range.
- the target movement of the mouth pot can prevent the spin of the lo-pot and the fall due to it while keeping the posture state quantity of the mouth pot in the target direction of the target movement stably. Can be determined. It is also necessary to determine a target motion that can prevent the movement trajectory of the mouth pot from deviating from the trajectory of the desired gait, or from impairing the continuous stability of the robot in one direction. it can. In addition, as long as the amount to be restricted falls within the allowable range, the amount to be restricted does not always need to be set to 0 or almost 0. Can be prevented.
- the first kinetic model may have a relatively low dynamic approximation accuracy
- the second kinetic model may have a higher dynamic approximation accuracy than the first kinetic model. Desirably.
- the difference between the first tentative instantaneous value and the second tentative instantaneous value of the target motion is determined by the state of the posture of the predetermined portion of the mouth port around the vertical axis or the floor normal axis. It is preferable to include a difference in the amount (fifth invention).
- the stability of the posture amount of the predetermined portion (for example, the upper body) in the one direction can be appropriately increased.
- the vertical component or the floor normal direction component of the moment of the floor reaction force substantially balanced with the instantaneous value of the target motion on the second dynamic model is used.
- the above-described forty-fourth invention can be easily realized by using, for example, the configuration of the thirty-fourth invention. That is, in the thirty-fourth aspect, at least the provisional instantaneous value of the desired floor reaction force is input to a third dynamic model representing a relationship between the movement of the lopot and the floor reaction force, and A third model calculating means for obtaining a third tentative instantaneous value of the target motion as an output of the dynamic model, wherein the means for determining a required value of the floor reaction force moment correction amount of the target floor reaction force is The required value is determined based on a difference between a target instantaneous value of the target exercise and a third provisional instantaneous value of the target exercise so that the difference approaches zero (the 47th invention).
- the third kinetic model and the third model calculating means correspond to the first kinetic model and the first model calculating means in the forty-fourth invention, respectively. Therefore, the forty-seventh invention is equivalent to the forty-fourth invention, and has the same effect as the forty-fourth invention.
- slip determination means is provided for determining occurrence of slip of the lopot operating following the target gait. JP2004 / 009472
- the permissible range setting means variably sets the permissible range according to the judgment result of the slip judging means (48th invention).
- the allowable range of the restriction target amount is variably set in accordance with the determination result of the actual occurrence of robot slippage, so that the occurrence of robot slippage can be reliably suppressed. If it is determined that slippage occurs, the allowable range should be set so as to narrow it.
- the occurrence of slip can be determined in the same manner as in the seventeenth to nineteenth inventions.
- the slip judging means judges the occurrence of slip based on at least the ground speed of the tip of the leg that is in contact with the ground (the 49th invention).
- the slip determination means obtains an apparent spring constant of the leg based on at least a temporal change rate of an actual floor reaction force acting on the grounded leg and a ground speed of the tip of the leg. Means for judging the occurrence of slip based on at least the apparent spring constant (50th invention).
- the slip judging means judges the occurrence of slip based on at least the actual floor reaction force acting on the grounded leg passed through a band pass filter having a frequency passing characteristic in a range near a predetermined frequency. (Invention 51).
- the occurrence of slip can be determined in the same manner as in the seventeenth to nineteenth inventions, respectively.
- the determination of the occurrence of slip can be made in any of the 49th to 51st inventions, but the determination of the occurrence of slip is made by combining two or more of the 49th to 51st inventions. You may do so.
- FIG. 1 shows a leg-type transfer in an embodiment of the present invention and a reference example related thereto.
- Fig. 2 schematically shows the overall configuration of a bipedal moving robot as a moving robot
- Fig. 2 shows the structure of the distal end of the robot leg in Fig. 1
- Fig. 3 shows the robot in Fig. 1.
- FIG. 4 is a block diagram showing a configuration of a control unit provided
- FIG. 4 is a block diagram showing a functional configuration of a control unit according to a reference example.
- Fig. 5 is a diagram for explaining the running gait generated in the embodiment and the reference example.
- Fig. 6 is a graph showing an example of the desired floor reaction force vertical component trajectory.
- Fig. 7 is the X component and Y of the desired ZMP trajectory.
- FIG. 8 is a diagram for explaining the upper body translation mode of the mouth pot
- FIG. 9 is a diagram for explaining the upper body tilt mode of the mouth pot
- FIG. 10 is a mouth.
- Fig. 11 (a) is a diagram for explaining the upper body single rotation mode of the pot
- Fig. 11 (a) is a diagram for explaining the anti-phase arm swing mode of the mouth pot in a plan view
- Fig. 11 (b) is a mouth.
- FIG. 12 is a diagram for explaining the anti-phase arm swing mode of the pot in a side view
- FIG. 12 is a diagram for explaining a dynamic model used in the embodiment.
- FIG. 13 is a flowchart showing the main routine processing of the gait generator in the reference example, FIG.
- FIG. 14 is a diagram for explaining the divergent state of the mouth pot
- FIG. 15 is a subroutine of S022 in FIG.
- FIG. 16 is a flowchart for explaining the process
- FIG. 16 is a diagram for explaining a normal gait and a supporting leg coordinate system
- FIG. 17 is a diagram illustrating a body trajectory of a normal gait and a supporting leg coordinate system
- Fig. 19 is a graph showing an example of setting the desired floor reaction force vertical component trajectory in a normal gait
- Fig. 20 is a floor reaction force horizontal component in a normal gait.
- FIG. 23 is a flowchart showing the subroutine processing of S024 in FIG. 13
- FIG. 24 is a flowchart showing the subroutine processing of S208 in FIG. 23
- FIG. 24 is a flowchart showing the subroutine processing of S306,
- FIG. 26 is a flowchart showing the subroutine processing of S412 in FIG.
- Figure 27 is acceptable
- FIG. 28 is a graph showing an example of the floor reaction force horizontal component considering the allowable range
- Fig. 29 is a graph showing an example of the body tilt angle acceleration
- Fig. 30 is a graph showing an example of the body tilt restoration moment ZMP converted value for restoring the robot's body tilt angle
- Fig. 31 shows the body tilt restoration moment ZMP converted value.
- Fig. 32 is a graph showing an example of the floor reaction force moment vertical component without considering the allowable range
- Fig. 33 is a floor reaction force moment vertical component considering the allowable range.
- FIG. 34 is a graph showing an example of the anti-phase arm swing moment
- FIG. 34 is a graph showing an example of the anti-phase arm swing moment
- FIG. 35 is a graph showing the anti-phase arm swing angular acceleration corresponding to the anti-phase arm swing moment of FIG. 34, and FIG. A graph showing an example of inverted phase arm swing restoration angular acceleration for restoring antiphase arm swing angle.
- Fig. 37 is a graph showing the antiphase arm swing recovery moment corresponding to the antiphase arm swing recovery angular acceleration in Fig. 36, and Fig. 38 is the vertical component of the floor reaction force moment in Fig. 33.
- 37 is a graph illustrating a vertical component of a floor reaction force moment obtained by combining the antiphase arm swing restoration moment of FIG.
- Fig. 39 is a flowchart showing the subroutine processing of S026 in Fig. 13, Fig.
- FIG. 40 is a graph showing an example of setting the floor reaction force horizontal component allowable range of the current time's gait, and Fig. 41 is the current time's step.
- Fig. 42 is a flow chart showing the subroutine processing of SO28 in Fig. 13 and
- Fig. 43 is a flow chart showing the subroutine processing of SO28 in Fig. 42.
- Fig. 44 is a graph showing an example of the tentative target ZMP 'of the current time's gait, the ZMP correction amount, and the corrected target ZMP, and Fig. 45 is a graph of S0300 in Fig. 13.
- FIG. 46 is a flowchart showing the subroutine processing, and FIG. 46 is a flowchart showing the subroutine processing of S11412 in FIG. Fig.
- FIG. 47 is a graph showing the relationship between the normal gait and the target gait's body position trajectory.
- Fig. 48 is a diagram showing another example of the body tilt mode (body tilt at the waist). Is a diagram for explaining another example of a dynamic model, and FIG. 50 is a vertical component of a desired floor reaction force in a walking gait.
- Fig. 51 shows the relationship between the vertical position of the body in the running gait of the robot and the vertical component of the floor reaction force, and
- Fig. 52 shows the relationship between the body vertical in the walking gait of the robot. It is a figure which shows the relationship between a direction position and a floor reaction force vertical component.
- FIG. 53 is a block diagram showing the functional configuration of the control unit according to the first embodiment of the present invention, and FIG.
- FIG. 54 is a flowchart showing the processing of the compensation total floor anti-camo horizontal component distributor shown in FIG. 53.
- FIG. 55 is a block diagram showing the processing of the model-operated floor anti-camo-motion vertical component determiner shown in FIG. 53
- FIG. 56 is a main routine processing of the gait generator in the first embodiment.
- FIG. 57 is a flowchart showing the subroutine processing of S3032 in FIG. 56
- FIG. 58 is a flowchart showing the subroutine processing of S31414 in FIG. It is.
- FIG. 59 is a block diagram showing a functional configuration of a control unit according to the second embodiment of the present invention
- FIG. 60 is a flowchart showing main routine processing of a gait generator in the second embodiment
- FIG. 60 is a flowchart showing the subroutine processing of S20334
- FIG. 62 is a flowchart showing the subroutine processing of S2114 in FIG.
- FIG. 63 is a block diagram illustrating a functional configuration of a gait generator according to the third embodiment of the present invention
- FIG. 64 is a diagram illustrating a setting example of a ZMP allowable range according to the third embodiment.
- FIG. 66 is a block diagram showing the processing of S 356 in FIG. 65
- FIG. 67 is a horizontal body position shown in FIG. Diagram for explaining the perturbation model
- Fig. 68 is a diagram for explaining the body posture angle correction perturbation model shown in Fig.
- FIG. 70 is a diagram for explaining the perturbation model.
- FIG. 70 is a block diagram showing processing of the antiphase arm swing angle correction perturbation model moment determination unit shown in FIG.
- FIG. 71 is a block diagram showing the processing of S3536 in FIG. 65 in the fourth embodiment of the present invention.
- FIG. 72 is a block diagram of the antiphase arm swing angle correction perturbation model moment determination unit shown in FIG. place
- FIG. 73 is a block diagram showing a functional configuration of a gait generator according to a fifth embodiment of the present invention
- FIG. 74 is a block diagram showing processing of a pseudo-order full model shown in FIG. FIG.
- FIG. 75 is a block diagram showing the processing of S355 in FIG. 65 in the sixth embodiment of the present invention.
- FIG. 76 is a block diagram showing a functional configuration of a control unit according to the seventh embodiment of the present invention.
- FIG. 77 is a flowchart showing main routine processing of the gait generator in the seventh embodiment. Is a flowchart showing the processing of S2334 in FIG. 77
- FIG. 79 is a flowchart showing the processing of the slip judging section shown in FIG. 76
- FIGS. 9 is a flowchart showing the subroutine processing of S 5 210, S 5 212, and S 5 214
- FIG. 83 shows the judgment result of the slip judging unit, the reduction rate of the allowable range, and the floor reaction force.
- FIG. 28 is a block diagram illustrating an example (eighth embodiment) of a modification of the process of the moment determination unit.
- a two-legged robot is taken as an example of a leg-type moving port.
- FIGS. 1 to 47 a description will be given of a reference example relating to a leg-type moving port control device of the present invention.
- An embodiment of the present invention to be described later has the same mechanical configuration as this reference example, and only a part of the gait generation processing and control processing of the mouth pot is different from the reference example. is there. Therefore, the description of this reference example is often referred to in the description of the embodiment described later.
- FIG. 1 is a schematic diagram showing an overall bipedal locomotion lopot as a legged locomotion lodge according to this reference example.
- a two-legged moving port (hereinafter referred to as a port) 1 is a pair of right and left legs (leg links) extending downward from an upper body (the base body of the robot 1) 2, 2 is provided.
- the two legs 2, 2 have the same structure, each having six joints.
- the six joints are for the rotation (rotation) of the crotch (lumbar region) (for rotation in one direction with respect to the upper body 3) in order from the upper body 3 side.
- the six joints are for the rotation (rotation) of the crotch (lumbar region) (for rotation in one direction with respect to the upper body 3) in order from the upper body 3 side.
- the symbols corresponding to the right leg and the left leg respectively. The same applies hereinafter
- the joint for rotation of the crotch (lumbar region) in the mouth direction about the X axis).
- each leg 2 At the lower part of the two joints of the ankle of each leg 2 1 8 R (L), 20 R (L). The foot that constitutes the tip of each leg 2 (foot) 2 2 R (L) At the same time, the top of both legs 2, 2 is connected via the three joints 1 OR (L), 12 R (L), and 14 R (L) of the crotch of each leg 2.
- the upper body 3 is attached. Inside the body 3, a control unit 60 described in detail later and the like are stored. In FIG. 1, the control unit 60 is shown outside the body 3 for convenience of illustration.
- the hip joint (or waist joint) ′ is composed of joints 1 OR (L), 12 R (L), and 14 R (L), and the knee joint is joint 16 R ( L), and the ankle joint is composed of joints 18 R (L) and 20 R (L).
- the hip and knee joints are linked by a thigh link 24 R (L). 2004/009472
- the knee joint and the ankle joint are connected by the lower leg link 26 R (L).
- a pair of left and right arms 5, 5 is attached to both upper sides of the upper body 3, and a head 4 is disposed at the upper end of the upper body 3.
- Each arm 5 has a shoulder joint composed of three joints 30 R (L), 32 R (L), and 34 R (L) and an elbow composed of joint 36 R (L). It has a joint, a wrist joint composed of a joint 38 R (L), and a hand portion 40 R (L) connected to the wrist joint. Since the head 4 has no direct relation to the gist of the present invention, a detailed description is omitted.
- the foot 2 2 R (L) of each leg 2 is given six degrees of freedom with respect to the upper body 3.
- 6 * 2 12 pieces including both legs 2 and 2
- * in this specification means multiplication as an operation for a scalar and multiplication as an operation for a vector.
- the desired motion of both feet 22R and 22L can be performed by driving the joint of (2) at an appropriate angle.
- the robot 1 can arbitrarily move in the three-dimensional space.
- each arm 5 can perform an exercise such as arm swing by rotating its shoulder joint, elbow joint, and wrist joint.
- a known 6-axis force sensor 50 is located below the ankle joint 18 R (L) and 20 R (L) of each leg 2 between the foot 22 and R (L). Are interposed.
- the 6-axis force sensor 50 is for detecting the presence / absence of landing of the foot 22 R (L) of each leg 2, the floor reaction force (ground contact load) acting on each leg 2, and the like.
- the detection signals of the three-directional components FX, Fy, Fz of the translational force of the floor reaction force and the three-directional components MX, My, Mz of the moment are output to the control unit 60.
- the upper body 3 includes the inclination angle and its angular velocity of the upper body 3 with respect to the Z axis (vertical direction (gravity direction)), and the rotation angle of the upper body 3 around the Z axis ( ⁇ one angle) and its angular velocity.
- An attitude sensor 54 is provided for detection, and a detection signal is output from the attitude sensor 54 to the control unit 60. It is.
- the attitude sensor 54 includes a three-axis acceleration sensor and a three-axis gyro sensor (not shown), and the detection signals of these sensors determine the attitude angle (tilt angle and angle) of the body 3 and its angular velocity. And is used to estimate the robot's self-position and orientation.
- each joint of the robot 1 has an electric motor 64 (see FIG. 3) for driving the joint, and a rotation amount of the electric motor (see FIG. 3).
- An encoder (mouth—tally encoder) 65 (see FIG. 3) for detecting the joint rotation angle) is provided, and a detection signal of the encoder 65 is output from the encoder 65 to the control unit 60.
- a joystick (operator) 73 is provided at an appropriate position of the mouth port 1, and the joystick 73 is operated.
- a request for the gait of the mouth port 1 can be input to the control unit 60 as needed, for example, by turning the lopot 1 moving straight. .
- FIG. 2 is a diagram schematically showing a basic configuration of a tip portion of each leg 2 (including each foot 22 R (L)) in the present reference example.
- a spring mechanism 70 is provided between the foot 22 and the 6-axis force sensor 50, and a sole (each foot 22) is provided.
- An elastic sole 71 made of rubber or the like is affixed to the bottom surface of R (L).
- the compliance mechanism 72 is constituted by the spring mechanism 70 and the sole elastic body 71.
- a rectangular guide member (not shown) attached to the upper surface of the foot 22 R (L) and an ankle joint 18 R (L) (FIG.
- FIG. 3 is a block diagram showing the configuration of the control unit 60.
- the control unit 60 is composed of a microcomputer and includes a first arithmetic unit 90 and a second arithmetic unit 92 composed of a CPU, an AZD converter 80, and a power counter 86. , A DZA converter 96, a RAM 84, a ROM 94, and a bus line 82 for exchanging data between them.
- the output signals of the 6-axis force sensor 50, posture sensor 54 (acceleration sensor and rate gyro sensor), joystick 73, etc. of each leg 2 are digitalized by the AZD converter 80.
- the data is sent to the RAM 84 via the bus line 82.
- the output of the encoder 65 (rotary encoder) of each joint of the mouth port 1 is input to the RAM 84 via the counter 86.
- the first arithmetic unit 90 generates a desired gait as described later and calculates a joint angle displacement command (a displacement angle of each joint or a command value of a rotation angle of each electric motor 64). , Sent to RAM 84. Also, the second arithmetic unit 92 reads the joint angle displacement command from the RAM 84 and the actual measured value of the joint angle detected based on the output signal of the encoder 65, and is necessary for driving each joint. The amount of operation is calculated and output to the electric motor 64 that drives each joint via the D / A converter 96 and the sample amplifier 64 a.
- FIG. 4 is a block diagram showing an overall functional configuration of a control device for a legged moving port according to this reference example.
- "Real robot" in Fig. 4 The other parts are constituted by the processing functions executed by the control unit 60 (mainly the functions of the first arithmetic unit 90 and the second arithmetic unit 92).
- the symbols R and L are omitted.
- the desired gait output by the gait generator 100 is a target body position / posture trajectory (trajectory of the target position and target posture of the body 3), a desired foot position / posture trajectory (the target of each foot 22) Orbit of position and target posture), target arm posture trajectory (trajectory of target posture of each arm 5), target total floor reaction force center point (target ZMP) trajectory, and target total floor reaction force trajectory.
- target body position / posture trajectory trajectory of the target position and target posture of the body 3
- a desired foot position / posture trajectory the target of each foot 22
- Orbit of position and target posture target arm posture trajectory
- target ZMP target total floor reaction force center point
- target ZMP target total floor reaction force trajectory
- the body posture in the gait means a pattern of time change (time-series pattern), and in the following description, may be referred to as “pattern” instead of “trajectory”.
- “posture” means spatial orientation. Specifically, for example, the body posture is the inclination angle of the body 3 in the roll direction (around the X axis) with respect to the Z axis (vertical axis) and the inclination angle and the yaw direction of the body 3 in the pitch direction (around the Y axis). The angle of rotation of the upper body 3 (around the Z axis). (Foot angle)-The foot posture is expressed by the spatial azimuth of two axes fixed to each foot 22 . In this specification, the body posture may be referred to as the body posture angle. Also, among the body postures, the posture in the vertical direction is sometimes referred to as the body posture inclination or the body posture inclination angle.
- the desired floor reaction force is generally expressed by an action point, a translation force and a moment acting on the point. Since the point of action is good for everywhere, countless expressions can be considered for the same desired floor reaction force.
- the target floor reaction force center point (the target position of the center point of the total floor reaction force) is used as the point of action to achieve the desired floor reaction force.
- the moment component of the desired floor reaction force is zero except for the vertical component (moment about the vertical axis (Z axis)).
- the horizontal component of the moment of the desired floor reaction force around the center point of the desired floor reaction force is zero.
- a desired gait is a set of a desired motion trajectory and a desired floor reaction force trajectory during one or more steps.
- the desired gait is the desired motion trajectory during one step and its ZMP trajectory. It is a pair with.
- the vertical position (body height) of the body 3 of the robot 1 is determined by the body height determination method previously proposed by the present applicant in Japanese Patent Application Laid-Open No. Hei 10-86080. Then, the translational floor anti-linear component is determined subordinately. Furthermore, the body horizontal position trajectory is determined so that the resultant component of inertia and gravity due to the motion of the desired gait has zero horizontal component of the moment generated around the target ZMP. This also determines the translational floor reaction force horizontal component. For this reason, in the specification of PCT Publication WO / 02Z 40224, only the target ZMP was sufficient as a physical quantity to be explicitly set with respect to the floor reaction force of the target gait.
- the floor reaction force vertical component (translational floor reaction force vertical component) is also important for control. Therefore, in the present invention, the trajectory of the robot 1, such as the desired vertical body position, is determined after explicitly setting the target trajectory of the floor reaction force vertical component. Therefore, in the present specification, the following b ′) is used as the definition of the desired gait in a narrow sense.
- a desired gait in a narrow sense is a set of a desired motion trajectory during one step, and a desired floor reaction force trajectory including at least the desired ZMP trajectory and the desired translational floor reaction force vertical component trajectory.
- the target gait will be used in the narrow sense of the above-mentioned target gait b ') unless otherwise specified, in order to facilitate understanding.
- “one step” of the target gait is used in the meaning from the time when one leg 2 of the robot 1 lands to the time when the other leg 2 lands.
- the target gait is used in the sense of a gait for one step, but it does not necessarily have to be for one step.
- a gait for a period of more than one step or shorter than one step (for example, half a step) may be used.
- the floor reaction force vertical component means “translational floor reaction force vertical component”
- the vertical component of the moment of the floor reaction force (the component around the vertical axis) Is distinguished from the term “floor reaction vertical component”.
- the floor reaction force horizontal component means “translational floor reaction horizontal component”.
- the period during which the mouth pot 1 supports its own weight with both legs 2 and 2 is not limited to the period when both legs are supported in the gait.
- the aerial period is the period during which both legs 2, 2 are off the floor (floating in the air).
- the leg 2 on the side that does not support the weight of the mouth pot 1 during the one-leg support period is called a “free leg”.
- there is no two-leg support period and the one-leg support period (landing period) and the aerial period are alternately repeated.
- both legs 2 and 2 do not support the weight of mouth port 1 during the aerial period, but legs 2 and the supporting legs, which were free legs during the one leg support period immediately before the aerial period, were not used.
- the two legs 2 are referred to as a free leg and a supporting leg, respectively, even in the aerial phase.
- the outline of the target gait generated by the gait generator 100 will be described using the running gait shown in FIG. 5 as an example. Other definitions and details regarding the gait are also described in Japanese Patent Application Laid-Open No. H10-86081 proposed earlier by the present applicant. 0 8 1. The contents that are not described in the official gazette are mainly explained.
- the running gait shown in FIG. 5 will be described.
- This running gait is similar to a normal running gait of a human.
- the foot 2 2 of only one of the left and right legs 2 (supporting leg) of the robot 1 lands (grounds), and the two legs 2 and 2 float in the air.
- the aerial period is repeated alternately.
- the first state in Fig. 5 is the state at the beginning (initial) of the one-leg support period, and the second state.
- the third state is at the beginning of the aerial phase following the one-leg support phase (at the end of the one-leg support phase), and the fourth state is the The state at the intermediate point, the fifth state, shows the state at the end of the aerial period (at the start of the next one-leg support period).
- the mouth pot 1 is the leg on the support leg (the leg 2 on the front side in the traveling direction of the mouth pot 1) at the start of the one-leg support period.
- the toe of the foot 22 on the supporting leg side (the foot 22 of the leg 2 on the rear side in the traveling direction of the mouth pot 1 in the third state in FIG. 5) Kicks the floor and jumps into the air. This ends the one-leg support period and starts the aerial period.
- the free leg in the one-leg support period exists at the back of the support leg at the beginning of the one-leg support period, as shown in the first state in Fig. 5. As shown in the state above, it is swung toward the front of the support leg toward the next scheduled landing position. Then, after the aerial period shown in the fourth state in Fig. 5, the mouth port 1 is the foot of the swing leg (leg 2 that had been a swing leg during the one-leg support period immediately before the start of the aerial period). 22 Landed on the heels of 2 and the next one leg support period begins.
- the basic outline of the target gait generated by the gait generator 100 will be described in consideration of the running gait of FIG. Although details will be described later, when the gait generator 100 generates the desired gait, the landing position / posture (scheduled landing position / posture) and the landing time (scheduled landing time) of the foot 22 on the swing leg side, etc.
- the basic required value (required parameter) for generating the desired gait is given to the gait generator 100 in accordance with the required operation of the joystick 73 or the like. Then, the gait generator 100 generates a desired gait using the required parameters.
- the gait generator 100 After determining the parameters that define some components of the desired gait, such as the desired foot position / posture trajectory of the desired gait, the desired floor reaction force vertical component trajectory, etc. The instantaneous value of the desired gait is sequentially determined using the gait parameters, and a time-series pattern of the desired gait is generated.
- the target foot position / posture trajectory (more specifically, the target trajectory of each spatial component (X-axis component, etc.) of the foot position and posture) is described in, for example, Japanese Patent No. 3233333. It is generated for each foot 22 using the finite time settling filter proposed in 450.
- This finite-time settling filter is a first-order lag filter with a variable time constant, that is, a filter whose transfer function is expressed in the form of 1 Z (1 + s). It is composed of multiple stages (three or more stages in this reference example) connected in series, and can generate and output a trajectory that reaches the specified value at the desired specified time. Things.
- the time constants of the unit filters in each stage are sequentially and variably set in accordance with the remaining time until the specified time from when the output of the finite-time settling filter is started. More specifically, as the remaining time becomes shorter, the value of ⁇ decreases from a predetermined initial value (> 0), and finally, at the specified time when the remaining time becomes 0, Is set to 0.
- the finite time settling filter is provided with a step input having a height corresponding to the specified value (more specifically, the amount of change from the initial value of the output of the finite time settling filter to the specified value).
- Such a limited-time settling filter not only generates an output that reaches a specified value at a specified time, but also sets the output change speed of the finite-time settling filter at the specified time to 0 or almost 0. Can be.
- the output change acceleration (differential value of the change speed) of the finite time settling filter can be set to zero or almost zero.
- the target foot position / posture trajectory generated by the finite-time setting file as described above is the target position / posture trajectory of each foot 22 in the support leg coordinate system described later fixed to the floor surface. .
- the target foot position / posture trajectory generated as described above shows that the position of each foot 22 gradually accelerates from its initial contact state (the state at the initial time of the target gait) to the expected landing position. Generated to start moving.
- the target foot position / posture trajectory finally gradually reduces the speed of change of the position to 0 or almost 0 by the scheduled landing time, and reaches the scheduled landing position at the scheduled landing time. Generated to stop. For this reason, the ground speed (the speed of change of the position of each foot 22 in the support leg coordinate system fixed to the floor) at the moment of landing of each foot 22 becomes zero or almost zero. Therefore, the landing impact is reduced even if the landing is performed from a state in which all the legs 2, 2 are in the air at the same time in the running gait (a state in the air).
- the vertical velocity of the upper body 3 becomes downward from the latter half of the aerial period due to the gravity acting on the robot 1, and remains downward even at the time of landing. Therefore, as described above, the desired foot position / posture trajectory is generated so that the ground speed at the moment of landing of each foot 22 becomes 0 or almost 0, and the dynamic equilibrium condition is satisfied as described later.
- the target position and attitude trajectory of the upper body 3 is generated, the relative speed of the foot 22 on the swing leg side to the upper body 3 becomes upward immediately before landing. That is, at the moment of the landing of the running gait, the target gait of the mouth pot 1 is a gait such that the leg 22 on the free leg side is retracted to the upper body 3 side.
- the robot 1 is viewed from the upper body 3 so that the ground speed of the foot 22 on the free leg side becomes 0 or almost 0 at the moment of landing. Raise foot 22 and land. As a result, the landing impact is reduced, and the landing impact is prevented from becoming excessive.
- the finite time setting filter is a series of three or more unit filters (for example, three steps) connected in series, so the speed of each foot 22 before the scheduled landing time ( Not only does the speed of change of the foot position) become 0 or almost 0, but also the acceleration of each foot 22 becomes 0 or almost 0 at the scheduled landing time and stops. That is, the ground acceleration at the moment of landing is also zero or almost zero.
- the landing impact is further reduced.
- the number of steps in the unit file of the finite time settling filter may be two, but in this case, The acceleration of each foot 22 at the scheduled landing time is generally not zero.
- the foot position trajectory was generated using a finite time settling filter.
- the change speed of the foot position at the scheduled landing time was 0 or almost 0.
- the target foot is calculated using a function such as a polynomial set so that the change acceleration of the foot position (the time differential value of the change speed) at the scheduled landing time becomes zero or almost zero.
- a position trajectory may be generated.
- the generation of the desired foot posture trajectory as described above, at the time when almost the entire bottom surface of each foot 22 is installed on the floor, the change speed of the posture of each foot 22, A function such as a polynomial is set so that the change acceleration becomes zero or almost zero.
- the desired floor reaction force vertical component trajectory is set, for example, as shown in FIG.
- the shape of the desired floor reaction force vertical component trajectory in the gait (specifically, the shape during the one-leg support period) is defined as a trapezoidal shape (a shape that is convex on the increasing side of the floor reaction force vertical component).
- the gait parameters floor reaction force vertical component trajectory parameters
- the desired floor reaction force vertical component is constantly set to zero.
- the desired floor reaction force vertical component trajectory should be set so that it is substantially continuous (so that the value does not become discontinuous).
- substantially continuous refers to the value jump that occurs when an analog continuous trajectory (real continuous trajectory) is digitally represented in a discrete time system. This does not mean that the continuity is lost.
- the target ZMP trajectory is set as follows. In the running gait shown in Fig. 5, as described above, the foot lands on the support leg side foot 22 and then kicks with the toe of the support leg side foot 22 and jumps up into the air. Landing on the heel of the side foot 2 2 Therefore, the target ZMP trajectory during the one-leg support period is, as shown in the upper part of Fig. 7, the heel of the support leg side foot 22 as the initial position, and then almost the entire bottom surface of the support leg side foot 22. Is set to move to the center in the front-rear direction of the foot 22 during the period in which the foot is in contact with the ground, and then to the toe of the support leg-side foot 22 by the time of leaving the floor.
- FIG. 7 shows the target ZMP trajectory in the X-axis direction (front-back direction), and the lower diagram in FIG. 7 shows the target ZMP trajectory in the Y-axis direction (left-right direction).
- the target ZMP trajectory in the Y-axis direction during the one-leg support period is set at the same position as the center position of the ankle joint of the supporting leg side leg 2 in the Y-axis direction, as shown in the lower part of Fig. .
- both legs 2, 2 are separated from the floor after the end of the one-leg support period, and the floor reaction force vertical component becomes zero.
- the target ZMP may be set to be discontinuous.
- the target ZMP is set so that it does not move from the target ZMP position at the time of leaving the bed (at the end of the one-leg support period).
- the target ZMP trajectory may be set so as to move in a (step-like) manner.
- this reference example as shown in the upper diagram of Fig.
- the position of the target ZMP trajectory in the X-axis direction during the aerial period is The toes were continuously moved from the toe to the landing position of the heel of the foot 22 on the free leg side.
- the position of the target ZMP trajectory in the Y-axis direction during the aerial period is the center of the ankle joint of the supporting leg
- the robot moves continuously from the Y-axis direction position to the Y-axis position at the center of the ankle joint of the swing leg side 'leg 2'.
- the target ZMP trajectory was made continuous (substantially continuous) throughout the gait.
- a desired gait is generated such that the moment of the resultant of the gravity and the inertial force around the target ZMP (excluding the vertical component) becomes zero (more specifically, the desired body position and posture). Adjust the trajectory). Taking into account the model's approximation error, the empty It is desirable to keep the target ZMP trajectory continuous (substantially continuous) even in the medium term.
- the horizontal component of the moment around the target ZMP is set to a certain value '(the value is 0 in this reference example, regardless of the position of the target ZMP. Since the desired gait such that it is not always 0 in the embodiment described later can be uniquely generated, it is not always necessary to keep the target ZMP continuous.
- the position and time of the break point of the target ZMP trajectory as shown in FIG. 7 are set as ZMP trajectory parameters (parameters defining the target ZMP trajectory).
- the meaning of “substantially continuous” of the ZMP trajectory is the same as in the case of the floor reaction force vertical component trajectory.
- the parameters of the ZMP orbit are determined so as to have a high stability margin and to prevent a sudden change.
- a state in which the target ZMP exists near the center of the smallest convex polygon (so-called supporting polygon) including the ground contact surface of the robot 1 is referred to as having a high stability margin (for details, see Japanese Patent Application Laid-Open No. H10-860). 81 No. 1).
- the target ZMP trajectory in Fig. 7 is set to satisfy such conditions.
- the target body position / posture, target foot position / posture, and a reference body posture described later are described in a global coordinate system.
- the global coordinate system is a coordinate system fixed to the floor. More specifically, a support leg coordinate system described later is used as the global coordinate system. '
- the gait generator 100 generates not only the desired body posture but also the reference body posture.
- the reference body posture is generated in accordance with the gait requirements (requests from the gait generator 100, such as a device such as an action planning unit or an external device (such as the joystick 73)). Posture.
- Target body posture (hereinafter referred to as the target body posture if no “reference” is attached) PC leak 004/009472
- the target body posture In walking, the target body posture should always always match the reference body posture as in the embodiment described in the specification of PCT Publication WO / 02/40224 by the present applicant.
- PCT Publication WOZ 02/40224 the concept of the reference body posture is not described, but since the target body posture pattern is given explicitly and preferentially, the target body posture It is the same thing that body postures always match.
- gait with an aerial period such as running.
- simply adjusting the horizontal acceleration of the upper body, etc. will result in the floor reaction force horizontal component and floor reaction force moment vertical component of the target gait.
- the dynamic equilibrium condition cannot be satisfied while the value is within the allowable range (or within the friction limit).
- the target body posture is intentionally shifted from the reference body posture as necessary. More specifically, by generating the operation modes described below in combination, the floor reaction force horizontal component and the floor reaction camo- ture vertical component of the desired gait fall within the allowable range (or within the friction limit). The kinetic equilibrium condition was satisfied while existing.
- a motion that changes the floor reaction force moment horizontal component and the floor reaction force horizontal component around the target ZMP without changing the floor reaction force vertical component is called a body motion mode.
- the vertical component of the floor reaction camouflage (the component around the vertical axis) is also perturbed, but we do not pay attention to this point.
- ⁇ ⁇ P and the change in the floor reaction force horizontal component per unit acceleration is A Fp.
- the mouth pot 1 when the body tilt angle acceleration (angular acceleration at the tilt angle of the body 3) is perturbed from a certain motion state of the robot 1 around a certain point Pr, the mouth pot 1
- the total center of gravity of the body does not perturb, but the angular momentum around the whole body center of gravity (excluding the component around the vertical axis) perturbs.
- the perturbation of the body tilt angle acceleration around the point Pr does not perturb the floor reaction force vertical component and the floor reaction force horizontal component, but perturbs the floor reaction force moment horizontal component around the target ZMP.
- the motion mode that perturbs the body tilt angle acceleration at mouth port 1 is called the body tilt mode.
- a motion that changes the floor reaction force moment horizontal component around the target ZMP without changing the floor reaction force vertical component and the floor reaction force horizontal component is called a body tilt mode.
- the perturbation of the angular acceleration of the body around the point Pq is the vertical floor reaction force moment around the target ZMP without perturbing the floor reaction force vertical component, floor reaction force horizontal component, and floor reaction force moment horizontal component. Perturb the components.
- the motion mode in which the robot 1 angular acceleration is perturbed in this way is called the body rotation mode.
- this motion mode is referred to as an anti-phase arm swing mode.
- the arm swing motion mode that perturbs the floor reaction force moment vertical component around the target ZMP without perturbing the floor reaction force vertical component, floor reaction force horizontal component, and floor reaction force moment horizontal component Called arm swing mode.
- FIGS 11 (a) and (b) show the state in which the antiphase arm swing angle is 0 az.
- AMaz is the change in floor component per unit angular acceleration per unit angular acceleration
- mu Fa is the change in floor reaction force horizontal component per unit angular acceleration.
- AFa is zero.
- the right arm 5R is accelerated forward and the left arm 5L is accelerated backward (shaking with angular acceleration / 3a> 0).
- the floor reaction force moment vertical component Maz acts in the direction of the arrow shown in (a) (positive direction of the vertical axis).
- a dynamic model of the robot 1 used in this reference example will be described.
- a simplified (approximate) dynamic model shown below is used.
- a kinematics model (a model that represents the structure and dimensions of joints and links, in other words, a model that represents the relationship between joint displacement and the position and orientation of the links) is also required. It is.
- FIG. 12 is a dynamic model of the mouth pot 1 used in the present reference example. As shown in the figure, this dynamic model is composed of two mass points 2 m and 2 m corresponding to each leg 2 of the mouth pot 1 and a mass point 3 m corresponding to the upper body 3, and a total of three mass points.
- This model consists of four flywheels, F Hx, F Hy, FHbz, and F Haz, which have one mass and have no mass. fly P
- Wheels F Hx, FHy, F Hbz, and FH az can rotate around X axis (front-rear axis), Y axis (left-right axis), Z axis (vertical axis), and Z axis (vertical axis), respectively. It is something.
- the relationship between the motion of the upper body 3 and the floor reaction force the relationship between the translational motion of the upper body 3 (upper body translation mode) and the floor reaction force, the tilting motion of the upper body 3 (upper body tilt mode) Between the floor reaction force and the upper body 3 (rotation mode) and the floor reaction force, as well as the arm swing movement (opposite phase) of the arms 5 and 5 (Arm swing mode) and the floor reaction force.
- the floor reaction force generated by the horizontal motion of the upper body mass 3m corresponds to the floor reaction force generated by the horizontal translational motion of the upper body 3 (body translation mode)
- the flywheel FHx and The floor reaction force generated by the rotary motion of FHy corresponds to the floor reaction force generated by the rotary motion of the upper body 3 at the tilt angle (body tilt mode).
- the rotational movement of the flywheel FHx corresponds to the rotational movement of the body 3 in the roll direction (around the X axis)
- the rotational movement of the flywheel FHy corresponds to the pitch direction of the body 3 inclining angle ( This corresponds to a rotational movement around the Y axis).
- the floor reaction force generated by the rotational motion of the flywheel F Hbz ′ corresponds to the floor reaction force generated by the one-rotational motion of the upper body 3 (the upper-body one-rotation mode).
- the floor reaction force generated by the rotational movement of the flywheel FHaz corresponds to the floor reaction force generated by anti-phase arm swing (anti-phase arm swing mode).
- each mass point 2m, 2m, 3m is a representative point of the corresponding part, or corresponds to a point that is uniquely determined geometrically from the position and orientation of that part.
- the position of the mass point 2 m of the supporting leg side leg 2 is a point located a predetermined distance above the representative point on the bottom surface of the foot 22 of the leg 2.
- Zb Vertical position of upper body mass point (Generally different from vertical position of upper body)
- Z Gtotal Vertical position of overall center of gravity
- Body mass point X position (The body mass position is a point offset from the point Pr by a predetermined distance in the forward and rearward direction of the body. The offset is the center of gravity of the strict model when standing upright, etc. (The position is determined so that the position of the center of gravity of this dynamic model is as close as possible. It is generally different from the horizontal position of the upper body.)
- ⁇ bx Body tilt angle around X axis with respect to vertical direction
- J Upper body moment of inertia (Equivalent moment of inertia in the upper body tilt mode; that is, the moment of inertia of FHx and FHy. In general, this does not match the moment of inertia of the three upper bodies of the actual robot 1. )
- J bz Body moment of inertia around the vertical axis (Equivalent moment of inertia in one rotation mode of the body. Inertia moments of the three body parts of the actual mouth pot 1 are generally It does not match.)
- J az Moment of inertia around the vertical axis of arm swing (Equivalent inertia moment in antiphase arm swing for spin cancellation. That is, FHz inertia moment.)
- Fx Floor reaction force X component (specifically, the longitudinal direction (X axis) component of the translational floor reaction force)
- Fy Floor reaction force Y component (specifically, the horizontal direction (Y axis) component of the translational floor reaction force)
- Fz Floor reaction force vertical component (Specifically, the vertical direction (Z axis) component of the translational floor reaction force. In this reference example, this is equal to the target translational floor reaction force vertical component.)
- Mx Floor reaction force moment X component around target ZMP (Details around floor reaction camoment—the longitudinal axis (X axis))
- each mass point 2 m and 3 m mean the position in the front-back direction (X-axis direction) and the position in the left-right direction (Y-axis direction).
- X-axis direction the position in the front-back direction
- Y-axis direction the position in the left-right direction
- the positional relationship with the location is determined in advance, and if one position is determined, the other position is uniquely determined. I have.
- the positional relationship between the body mass point 3 m and the position of the body 3 is determined by the posture angle of the body 3 (hereinafter, the posture angle Means the angle of inclination and the angle of one angle.) Is determined in advance, and if one position and attitude angle are determined, the other position is uniquely determined. .
- dX / dt represents the first derivative of X
- d2X / dt2 represents the second derivative of X. Therefore, if the variable X is displacement, dX / dt means velocity and d2X / dt2 means acceleration. g indicates the gravitational acceleration constant. Here, g is a positive value.
- Equations 01, 02x, 02y, 03x, 03y, and 03z The equations of motion of the above dynamic model (expressions representing the dynamic equilibrium conditions) are represented by Equations 01, 02x, 02y, 03x, 03y, and 03z.
- F z mb * (g + d2Zb / dt2) + msup * (g + d2Zsup / dt2)
- Mx mb * (Yb-Yzmp) * (+ d2Zb / dt2)
- M z mb * (Xb-Xzmp) * (d2Yb / dt2)-mb * (Y-Yzmp) * (d2Xb / dt2)
- Equation 0 3 z Further, the following relational expression holds for the position of the center of gravity of the entire lopot.
- Y Gtotal (mb * Yb + msup * Ysup + mswg * Yswg) / mtotal
- the ⁇ Fp is a perturbation amount of Fx or Fy when d2Xb / dt2 or d2Yb / dt2 is perturbed by a unit amount in the expression 0 2 X or the expression 0 2 y, and is thus obtained by the following expression.
- the change AFp of the floor reaction force horizontal component per unit acceleration in each horizontal axis (X axis, Y axis) in the body translation mode is equivalent to the mass of the body mass point 3m in the above dynamic model.
- ⁇ is the perturbation amount of My or Mx when d2Xb / dt2 or d2Yb / dt2 is perturbed by a unit amount in Expression 03y or Expression03X, it is obtained by the following expression.
- the change ⁇ of the floor reaction force moment horizontal component per unit acceleration in each horizontal axis (X-axis, Y-axis) direction of the upper body translation mode is calculated by adding the upper body mass of 3m It is the product of the height (vertical position) from the target ZMP.
- the relationship between the position of the upper body mass point 3m and the target ZMP and the movement of the upper body mass point 3m is as follows: the upper body mass point 3m corresponds to the inverted pendulum mass point, and the target ZMP corresponds to the inverted pendulum fulcrum.
- ⁇ in the Y-axis direction is, more accurately, the sign of the right side of Expression 07 inverted.
- a Mbz is the perturbation amount of Mz when d20bz / dt2 is perturbed by a unit amount in Expression 03z, and is obtained by the following expression.
- the change A Mbz of the floor reaction force moment component per unit acceleration in the upper body rotation mode corresponds to the moment of inertia of the flywheel F Hbz corresponding to the upper body rotation.
- a Maz is a perturbation amount of Mz when d20 az / dt2 is perturbed by a unit amount in Expression 03z, and thus is obtained by the following expression.
- the change A Maz of the floor reaction force moment component per unit angular acceleration of antiphase arm swing corresponds to the inertia moment of the flywheel FHaz corresponding to arm swing.
- the gait generator 100 in the present reference example is a target gait for one step from the landing of one leg 2 of the robot 1 to the landing of the other leg 2 (in the narrow sense described above).
- the desired gait for one step is generated in order with the desired gait) as a unit. Therefore, in the running gait of FIG.
- target gaits from the start of the one-leg support period to the end of the following aerial period (at the start of the next one-leg support period) are generated in order.
- the target gait to be newly generated is “this gait”
- the next target gait is “next gait”
- the next target gait is “next next gait”
- the target gait generated just before the “current gait” is called the “previous gait”.
- the gait generator 100 When the gait generator 100 newly generates the current time's gait, the gait generator 100 lands on the free leg side foot 22 up to two steps ahead of the mouth port 1.
- the required values (requests) of the planned position / posture and the scheduled landing time are input as the required parameters for the gait (or the gait generator 100 reads the required parameters from the storage device). Then, the gait generator 100 uses these required parameters to obtain a desired body position / posture trajectory, a desired foot position / posture trajectory, a desired ZMP trajectory, a desired floor reaction force vertical component trajectory, a desired arm posture trajectory, etc. Generate At this time, part of the gait parameters that define these trajectories is modified as appropriate to ensure continuity of walking.
- FIG. 13 is a flowchart (structured flowchart) showing a main routine of a gait generation process executed by the gait generator 100. '
- various initialization operations such as initializing time t to 0 in S 0 10 are performed. This process is performed when the gait generator 100 is activated. Next, the process proceeds to SO 14 via SO 12, and the gait generator 100 waits for a timer interrupt for each control cycle (the arithmetic processing cycle of the flowchart in FIG. 13).
- the control cycle is ⁇ 1;
- gait switching point means a timing at which the generation of the previous time's gait has been completed and the generation of the current time's gait has started. The control cycle becomes the gait switching point.
- the request parameter given to the gait generator 100 from the joystick 73 or the like is the landing position / posture (foot 2) of the free leg side foot 2 2 up to two steps ahead. 2) After landing, the foot position and posture when the sole is rotated without slipping so that the sole touches the floor almost completely) and the required values for the scheduled landing time are included.
- the required value of the gait and the required value of the second step are assumed to correspond to the current time's gait-the next time's gait, respectively, before the generation of the current time's gait (switching of the gait in S016). It is given to the gait generator 100. These required values can be changed even during the generation of the current time's gait.
- the system is determined.
- the expected landing position of the free leg side foot 2 2 (22 L in the figure) related to the current time's gait (the first step).
- the Y-axis direction (the left and right direction of the foot 22 on the supporting leg side of the gait). It is assumed that the position and orientation have been moved by xnext and ynext and rotated by 0 znext around the Z axis (around the vertical axis).
- the supporting leg coordinate system is such that the supporting leg side foot 22 is in a horizontal posture (more generally, a posture parallel to the floor surface) and almost the entire bottom surface of the supporting leg side foot 22 is on the floor.
- the representative point of the foot 22 coincides with the origin, and the horizontal plane passing through the origin is the XY plane.
- System Coordinat system fixed to the floor).
- the X-axis direction and the Y-axis direction are the front-rear direction and the left-right direction of the support leg side foot 22, respectively.
- the origin of the support leg coordinate system is not necessarily the representative point of the foot 22 when the entire bottom surface of the foot 22 is in contact with the floor (the position of the foot 22). It is not necessary to match with the representative point, and a point on the floor different from the representative point may be set.
- the next time's gait support leg coordinate system is, as shown in the figure, when the foot 22 is landed in accordance with the required value of the scheduled landing position and posture of the free leg 22 L of the current time's gait (foot 22.
- a point (more specifically, a point on the floor that matches the representative point) is defined as the origin, and the front and rearward and left and right directions of the foot 22 L in the horizontal plane passing through the origin are respectively defined as the X'-axis direction and the Y'-axis direction. Is a coordinate system.
- the gait support leg coordinate system (see X, Y coordinates in Fig. 16) is determined according to the required value of the expected landing position and posture of the free leg side foot 22 of the second step. .
- the gait cycle of this time is based on the scheduled landing time of the foot 22 on the supporting leg side of the current time gait (required value), and the landing time of the foot 22 on the free leg side of the first step (current gait). Required value), and the next gait cycle is the first step 2
- the required parameter is input to the gait generator 100 by operating the joystick 73 at the point g.
- the required parameter or the corresponding The position and orientation of the supporting leg coordinate system and the gait cycle may be stored as the movement schedule of the robot 1.
- a command (request) from a control device such as a joystick 73 and the movement history of one of the mouth pots up to that time, the next and next gait support leg coordinate system, and The next gait cycle may be determined.
- the gait generator 100 determines the gait parameters of the normal turning gait as a virtual periodic gait following the current time's gait.
- the gait parameters are a foot trajectory parameter defining a desired foot position / posture trajectory in a normal turning gait, a reference body posture trajectory parameter defining a reference body posture trajectory, and a reference arm posture trajectory.
- a reference arm posture trajectory parameter that defines the parameters
- the ZMP trajectory parameter that specifies the target ZMP trajectory
- the floor reaction force vertical component trajectory parameter that specifies the target floor reaction force vertical component trajectory.
- parameters defining the allowable range of the floor reaction force horizontal component and the allowable range of the floor reaction force moment vertical component are also included in the gait parameters.
- steady turning gait refers to the motion state (foot) of the mouth port 1 at the gait boundary (in this reference example, the gait boundary for each step) when the gait is repeated. It is used to mean a periodic gait in which discontinuity does not occur in the flat posture, body position and posture.
- normal turning gait may be abbreviated as “normal gait J”.
- the normal turning gait which is a periodic gait
- the gait composed of the second turning gait is a gait for one cycle of the normal turning gait, and the gait for one cycle is repeated.
- the term “turn” is used because, when the turning rate is zero, it means straight traveling, so that straight traveling can be included in turning in a broad sense.
- the generated gait is the running gait of FIG. 5 described above
- the current time's gait of the desired gait is a running gait having a one-leg support period and an aerial period.
- the first turning gait and the second turning gait are both gaits having a one-leg support period and an aerial period, similarly to the current time gait. That is, the basic gaits of the first turning gait and the second turning gait are the same as the current time gait.
- one cycle of the normal turning gait requires at least two gaits in the narrow sense described above. It is also possible to set a complex stationary turning gait in which gaits of three or more steps are gaits for one cycle.
- the normal turning gait is used only to determine the divergent component (details will be described later) at the end (end time) of the current time's gait, as described later. For this reason, using a normal turning gait with a gait of three or more steps as one cycle has little effect despite the complicated gait generation processing. Therefore, the gait for one cycle of the normal turning gait in this reference example is composed of two gaits (first and second turning gaits).
- a normal turning gait composed of a plurality of gaits in a narrow sense is regarded as one gait.
- the normal turning gait is provisional for the gait generator 100 to determine the divergent component at the end of the current time's gait, the body vertical position speed, the body posture angle, and the motion state of the robot 1 such as its angular velocity.
- the gait generator 100 does not output the gait generator 100 as it is.
- “divergence” means that the position of the upper body 3 of the bipedal moving robot 1 is shifted to a position far away from the position of both feet 22 and 22 as shown in FIG. .
- the value of the divergent component means that the position of the upper body 3 of the biped locomotion robot 1 is the position of both feet 22 and 22 (more specifically, the position of the foot 22 on the supporting leg side is set to It is a numerical value that represents how far away from the global coordinate system (the origin of the support leg coordinate system).
- the gait is generated using the divergent component as an index so that the desired gait is continuously generated without the divergence.
- a normal gait which is a typical example of a continuous gait (a periodic gait in which the gait of the same pattern can be repeated without generating a discontinuity in the gait trajectory, Even if the initial divergent component of a gait that does not diverge even after infinite repetition) (the divergent component at the initial time of the normal gait), it is not simply 0, but if the parameters of the normal gait change, Its initial divergent component also changes. In other words, the appropriate divergent component changes depending on the form of gait such as how to walk or run.
- the normal gait following the current gait to be generated is set according to the required parameters related to the current gait, and the initial divergent component of the normal gait is obtained. From this, the current gait is generated such that the terminal divergent component of the current time gait matches the initial divergent component of the normal gait (more generally, the current time's gait is made continuous or close to the normal gait) .
- the basic guidelines for generating such a gait are the same as those of the PCT publication WO / 02/40224 previously proposed by the present applicant.
- Equation 1 1 Equation 1 1 where the horizontal position of the upper body mass represents the horizontal position Xb of the upper body mass in the dynamic model shown in FIG.
- ⁇ and ⁇ ′ are certain predetermined values. These ⁇ 0 and ⁇ 'values are almost the same, but they are not exactly the same. Then, during running, it is necessary to slightly change the value at the time of generating the walking gait of PCT publication WOZ02 / 40224.
- the foot trajectory of the normal gait parameters is set so that the foot position / posture trajectory is connected in the order of the current gait, the first turning gait, and the second turning gait.
- the parameters are determined.
- a specific setting method will be described with reference to FIG.
- the foot 22 of the leg 2 on the supporting leg side is referred to as a supporting leg foot
- the foot 22 of the leg 2 on the free leg side is referred to as a free leg foot.
- the “initial” and “end” of a gait mean the start time and end time of the gait, respectively, or the instantaneous gait at those times.
- the foot trajectory parameters consist of the position and orientation of the supporting leg foot and the free leg foot at the beginning and end of the first turning gait and the second turning gait, the gait cycle of each turning gait, etc. Is done.
- the first swing gait initial free leg foot position / posture is the current gait end support leg foot position / posture viewed from the next time gait support leg coordinate system. In this case, in the running gait, the support leg foot 22 at the end of the gait this time is moving into the air. Then, the current gait end support leg foot position / posture is calculated from the initial gait initial support leg foot position / posture (-previous gait end free leg foot position / posture) for the second step in the above-mentioned required parameters.
- required value of expected landing position / posture of free leg side foot 22 (required value of expected landing position / posture in next time gait of supporting leg foot 22 of current gait) or next time corresponding to the required value
- the foot position / posture trajectory (the trajectory viewed from the next time's gait support leg coordinate system) leading to the next gait end swing leg foot position / posture determined according to the gait support leg coordinate system It is obtained by generating using a finite-time settling filter.
- the foot 22 is rotated in the pitch direction by a predetermined angle to the horizontal position so that the toe is lowered while keeping the foot 22 in contact with the ground.
- the position and posture of the foot when the gait supports the next time gait It is determined to match the position and orientation of the leg coordinate system.
- the foot position and posture of the next step at the end of the gait are calculated based on the required value of the landing position and posture of the free leg side foot 22 of the second step in the required parameter.
- the first turning gait initial support leg foot position / posture is the current gait end free leg foot position / posture viewed from the next time gait support leg coordinate system.
- the next time gait end free leg foot position / posture is one of the next time gait support leg coordinate system or one of the required parameters corresponding thereto. It is determined according to the required value of the expected free leg landing position / posture of the step (this time's gait). That is, from the position and posture, the foot 22 is rotated from the position and posture to lower the toe while keeping the foot 22 in contact with the ground, and the foot 22 is rotated by lowering the toe.
- the representative point of the foot when almost the entire bottom surface is brought into contact with the floor surface is determined so as to coincide with the origin of the next time's gait support leg coordinate system.
- the first turning gait end free leg foot position / posture is calculated in the same way as the method for determining the current gait end free leg foot position / posture and the next time gait end free leg foot position / posture. Is determined based on the position and orientation of the next-time gait support leg coordinate system viewed from the user. More specifically, the foot position and posture of the first swing gait end free leg foot are set at a predetermined angle from the position to the horizontal position so that the foot 22 does not slip while the foot 22 is grounded. The foot position and orientation when rotated are set to match the position and orientation of the next-time gait support leg coordinate system viewed from the next-time gait support leg coordinate system.
- the supporting leg foot 22 is off the floor and in the air.
- Support Leg foot 2 First turn gait support
- the planned landing position / posture of the legs / foot is set.
- the first turning gait support leg foot landing scheduled position / posture is set based on the position / posture of the next / next gait support leg coordinate system viewed from the next / time gait support leg coordinate system. More specifically, the first turning gait support leg foot landing scheduled position / posture is the position / posture of the next next gait support leg coordinate system viewed from the next time gait support leg coordinate system.
- the next-next gait support leg coordinate system is based on the relative position and orientation of the next-next gait support leg coordinate system and the next-next gait support leg coordinate system. It is set to match the relative position and orientation relationship with the supporting leg coordinate system.
- the first turning gait end support leg foot position / posture is calculated from the first turning gait initial support leg foot position / posture in the same manner as when the first turning gait initial support leg foot position / posture is obtained.
- the foot position and posture trajectory (or more specifically, the trajectory viewed from the next time's gait support leg coordinate system) leading to the first turning gait support leg foot landing scheduled position / posture are converted to the finite time settling filter up to the end of the first turning gait. It is obtained by generating using
- the initial swing leg foot position / posture of the second turning gait is the foot position / posture of the first turning gait end supporting leg as viewed from the next-time gait support leg coordinate system.
- the initial support leg foot position / posture of the second turning gait is the first turning gait end free leg foot position / posture viewed from the next-time gait support leg coordinate system.
- the second turning gait end free leg foot position / posture is the current gait end free leg foot position / posture viewed from the gait support leg coordinate system.
- the second 'turning gait end support leg foot position / posture is the current gait end support leg foot position / posture viewed from the current gait support leg coordinate system.
- the gait cycle of the first turning gait and the second turning gait is set to be the same as the next time gait cycle.
- the gait cycles of the first turning gait and the second turning gait are not necessarily required to be the same as each other, but it is important that each cycle is determined at least according to the next gait cycle. preferable.
- this time gait, 2 is not necessarily required to be the same as each other, but it is important that each cycle is determined at least according to the next gait cycle. preferable.
- Other exercise parameters including time parameters such as the two-leg support time) of the first turning gait and the second turning gait are determined according to the gait conditions ( Determine whether the speed of the event is within the allowable range, does not exceed the movable angle, does not interfere with the floor, etc.).
- a reference body posture trajectory parameter that defines a reference body posture trajectory to be followed by the target body posture is determined.
- the reference body posture is connected between the beginning of the normal gait (the beginning of the first turning gait) and the end (the end of the second turning gait) (the reference body posture at the beginning and end of the normal gait). It is not necessary to maintain a constant posture as long as it is set so that the posture angle and its angular velocity coincide with each other.
- the inclination angle vertical The attitude related to the angle of inclination with respect to the direction
- the one corner trajectory of the reference body posture (hereinafter, also referred to as the reference one corner trajectory) 0 bz may be, for example, a constant angular velocity (average turning speed of a normal gait).
- the trajectory may be sinusoidal as in the example of the reference antiphase arm swing trajectory described later (FIG. 18).
- the reference angle and its angular velocity are set to be continuous.
- the angle angle trajectory of the target body posture (hereinafter, the angle angle trajectory of the target ) Shall be made to coincide with the reference unilateral orbit.
- the process proceeds to S104, and the reference arm posture trajectory parameter is determined. Specifically, the position of the center of gravity of the entire arms 5, 5 (the position of the center of gravity relative to the upper body 3), the distance between the left and right hands (tips of the arms 5, 5) in the left and right direction, The parameters related to the antiphase arm swing angle are determined.
- Reference out-of-phase arm swing The corner may be set as shown in FIG. 18 when turning left as shown in FIG. 17, for example.
- the reference antiphase arm swing angle 0 azref is opposite phase at the boundary of the gait (the end of the second gait and the boundary of the next first gait) when the normal gait is repeated.
- the arm swing angle and angular velocity are both continuous, and the relative relationship between the support leg and the antiphase arm swing angle at the beginning of the first turning gait is It is set to match the relative relationship. That is, the initial anti-phase arm swing angular velocity of the first turning gait matches the terminal anti-phase arm swing angular velocity of the second turning gait, and the terminal anti-phase arm swing angle of the second turning gait is the normal gait turning.
- the angle (sum of the turning angles of the first turning gait and the second turning gait) plus the initial antiphase arm swing angle of the first turning gait is set.
- the reference antiphase arm swing angle 0 azref is a sinusoidal waveform in FIG. 18, it may be set to a constant angular velocity. Alternatively, the average value of the angle of the support leg and the angle of the swing leg may be used.
- the center of gravity (the relative position with respect to the upper body 3) of the entire arms 5 and 5 in the target arm posture is set so as to be kept constant with respect to the upper body 3.
- a floor reaction force vertical component trajectory parameter is set.
- the floor reaction force vertical component trajectory specified by the parameter reaction is set to the first turning gait and the The floor reaction force vertical component trajectory parameters are set so that in any of the two turning gaits, as shown in Fig. 6, the gait is substantially continuous (values do not fly in steps). That is, the desired floor reaction force vertical component trajectory of the normal turning gait is set in a pattern as shown in Fig. 19. In this pattern, in both the first turning gait and the second turning gait, the floor reaction force vertical component changes to a trapezoidal shape during the one-leg support period, and the floor reaction force vertical component is maintained at 0 during the airborne period. .
- the time of the break point of this pattern and the height (peak value) of the trapezoid are set as the parameters of the floor reaction force vertical component orbit.
- the total gait of the floor reaction force vertical component (the normal gait for the period of both the first turning gait and the second turning gait)
- the average value during one cycle of the period is matched with the weight of the robot. That is, the average value of the vertical component of the floor reaction force is set to be the same as the gravity acting on the mouth port 1 and in the opposite direction.
- the conditions of the normal gait are all the state variables of the gait (the position of each part of the mouth port 1, the position of each part of the gait, as viewed from the supporting leg coordinate system (the coordinate system set on the ground contact surface of the supporting leg side foot 22)).
- Posture speed, etc.
- the next supporting leg coordinate system supporting leg coordinate system of the next first turning gait
- the end state of the second turning gait hereinafter, this condition is sometimes referred to as the boundary condition of the normal gait).
- the difference between the overall center-of-gravity vertical speed at the end of the normal gait and the overall center-of-gravity vertical speed at the beginning of the normal gait (specifically, the overall center-of-gravity vertical speed at the end of the second gait and the first turning) (The difference from the vertical velocity of the entire center of gravity at the beginning of the gait) must also be zero. Since the above difference is the integral of the difference between the ground reaction force vertical component and gravity (first-order integration value), it is necessary to set the floor reaction force vertical component trajectory as described above to make the difference zero. There is.
- the average value of the floor reaction force vertical component in each of the first turning gait and the second turning gait is set to match the own weight of the lopot 1. More specifically, for example, after setting the time of the break point of the trapezoidal portion of the floor reaction force vertical component trajectory in each turning gait according to the gait cycle of the first turning gait and the second turning gait, The height of the trapezoidal portion was determined so that the average value of the floor reaction force vertical component in each of the first turning gait and the second turning gait coincided with the robot 1's own weight (the height of the trapezoid). The height of the trapezoid is determined by solving the above equation as an unknown and solving the equation representing the condition for matching the average value and the own weight).
- the difference between the overall center-of-gravity vertical speed at the end of the first turning gait and the overall center-of-gravity vertical speed at the beginning of the first turning gait is also zero.
- the difference from the vertical velocity of the entire center of gravity at the beginning of the turning gait is also zero.
- this is not necessary. For example, if the upper body vertical position is too high or too low near the boundary between the first turning gait and the second turning gait, and it is likely to be in an unreasonable posture, then in each turning gait
- the height and the like of the trapezoid of the floor reaction force vertical component trajectory of each turning gait may be corrected from the state where the average value and the own weight are matched.
- the allowable range of the floor reaction force horizontal component [Fxmin, Fxmax] (more specifically, Are defined as shown in Fig. 20.
- the negative line in Fig. 20 indicates the floor reaction force horizontal component allowable lower limit Fxmin, and the positive line indicates the floor reaction force horizontal component allowable upper limit Fxmax.
- the floor reaction force horizontal component is generated by the friction between the floor and the foot 2 2, but the friction cannot be generated as much as there is a limit. Therefore, the floor reaction force horizontal component of the desired gait must always be within the friction limit in order to prevent slipping when the actual lopot 1 moves according to the generated desired gait.
- the floor reaction force horizontal component allowable range is set, and as described later, the target gait is set so that the floor reaction force horizontal component of the target gait falls within this allowable range.
- Fxmin must always be set to be equal to or more than the * floor reaction force vertical component
- Fxmax must be set to * the floor reaction force vertical component or less.
- the simplest setting method is the following formula. Where ka is a positive constant smaller than 1.
- Fxmin — ka * 2 * Floor reaction force vertical component
- the floor reaction force horizontal component permissible range in Fig. 20 is an example set in accordance with Eq.
- the value and time at the break point such as the trapezoidal waveform in Fig. 20 may be set as the parameters defining the floor reaction force horizontal component allowable range.
- the value of (ka *) in Equation 12 may be simply set as a parameter.
- the allowable range [Mzmin, Mzmax] Specifically, the parameters that define this are set as shown in Figure 21.
- the polygonal line on the negative side in Fig. 21 represents the floor reaction force moment vertical component allowable lower limit Mzmin, and the polygonal line on the positive side represents the floor reaction force moment vertical component allowable upper limit Mzmax.
- the vertical component of the floor reaction force moment is generated by friction between the floor and the foot 22, but the friction cannot be generated as much as possible, and there is a limit. Therefore, in order to avoid spinning when the actual lopot 1 moves according to the generated target gait, the floor component of the target gait must be within the friction limit at all times. No. Therefore, in order to satisfy this condition, the floor reaction force moment vertical component allowable range is set, and as described later, the floor reaction force moment vertical component of the target gait is set within this allowable range. A desired gait was generated.
- Mzxmin —ka * * r * Floor reaction force vertical component
- the allowable range of the floor anti-camouflage vertical component in FIG. 21 is an example set according to the equation 1 0 1 2.
- the value and time at the break point such as the trapezoidal waveform in Fig. 21 may be set.
- r is preferably calculated at each moment from the target ZMP and the ground contact surface, but may be a constant.
- a ZMP trajectory parameter that defines the ZMP trajectory of the normal gait combining the first turning gait and the second turning gait is set.
- the target ZMP trajectory is set so as to have a high stability margin and not to change abruptly as described above.
- the target ZMP must be within the ground contact. Therefore, in this reference example, the position of the target ZMP in the X-axis direction of each of the first turning gait and the second turning gait of the normal gait is, as shown in the upper diagram of FIG.
- the heel of the flat 22 is set as an initial position, and is set so as to remain at that position until almost the entire bottom surface of the foot 22 is in contact with the ground.
- the target ZMP moves to the center of the supporting leg foot 22 and moves to the toe until the foot 22 comes into contact with the toe, and then toes of the supporting leg foot 22 until leaving the floor. Is set to stay at After that, as described above, the target ZMP is from the toe of the supporting leg foot 22 to the heel of the free leg foot 22 before the landing of the next free leg foot 22 as described above. It is set to move continuously to the position.
- the target ZMP trajectory (trajectory in the X-axis direction) of the normal gait consisting of the first gait and the second gait is as shown in FIG.
- the time and position of the turning point of the target ZMP trajectory are set as ZMP trajectory parameters.
- the time of the break point is set according to the gait cycle of the first turning gait and the second turning gait determined according to the required parameter, and the position of the break point is determined by the next time the gait is supported.
- the position of the .ZMP track in the Y-axis direction is set in the same manner as that shown in the lower diagram of FIG. More specifically, the trajectory of the target ZMP in the Y-axis direction in the first turning gait is set in the same pattern as that in the lower diagram of FIG. 7, and the trajectory of the target ZMP in the Y-axis direction in the second turning gait is The trajectory is the same as that of the first turning gait, and is connected to the end of the trajectory.
- a normal gait must be a gait in which the state variables at the beginning and end are continuously connected.
- the definition of the normal gait is different from the definition of the gait in the narrow sense described above until the determination of the normal gait.
- the termination and period are set as shown in Figure 19 for convenience. That is, the time at which the floor reaction force vertical component has decreased to some extent in the latter half of the one-leg support period of the first turning gait is set as the initial time Ts of the normal gait.
- the initial time Ts is, as shown in FIG. 7, the moment when the bottom surface of the support leg foot 22 is almost grounded and the moment immediately before it changes to toe grounding (the sole grounding period in FIG. 7).
- the cycle Tcyc of the normal gait is equal to the first turning gait. 2004/009472
- Te is the end time of the normal gait. Te is set to the time obtained by adding Tcyc to Ts.
- the body posture angle and the antiphase arm swing angle should be continuous at the boundary of the gait.
- the initial body posture angular velocity of the normal gait and the terminal body posture angular velocity must match, and the initial antiphase arm swing angular velocity and the terminal antiphase arm swing angular velocity of the normal gait must match.
- the above-mentioned period is a period for adjusting the body posture angle trajectory and the anti-phase arm swing angle trajectory for doing so.
- a time Tm at which the second turning gait becomes the second turning gait after the aerial period of the first turning gait from the initial time Ts and the floor reaction force vertical component has increased to a predetermined magnitude is set.
- a time Ts2 at which the floor reaction vertical component decreases to some extent in the latter half of the one-leg support period of the second turning gait is set.
- the time Tm2 at which the first turning gait becomes the first turning gait after the aerial period of the second turning gait and the vertical component of the floor reaction force increases to a predetermined magnitude is set.
- the time Tm is desirably set at the moment when almost the entire bottom surface of the support leg foot 22 is touched or immediately thereafter.
- the time Ts2 is the same as the initial time Ts at the moment when the foot 22 is almost completely grounded from the ground state to the toe ground state. Or it is desirable to set immediately before.
- the target ZMP of FIG. 22 set in S110 of FIG. 15 and these times Tm, Ts2, and Tm2 will be described.
- the target ZMP is With the heel of leg foot 22 as the initial position, it stays at that position until almost the entire bottom surface of foot 22 is in contact with the ground, and then the moment when it begins to move to the center of support leg foot 22 is time Tm. After that, it is desirable that the moment when the target ZMP completes its movement to the toe until the toe of the supporting leg foot 22 is in contact with the ground is time Ts2.
- the target ZMP stays at the heel of the support leg foot 22 as an initial position until almost the entire bottom surface of the foot 22 touches the ground, and then the support leg It is desirable that the moment at which the foot 22 begins to move to the center be set to time Tm2.
- the reason for setting as described above will be described later.
- the period for restoring (adjusting) the body posture angle and the period for restoring (adjusting) the antiphase arm swing angle may be set separately.
- the initial state of the normal gait is calculated.
- the initial state calculated here is the initial body horizontal position speed of the normal gait (the initial body position and the initial body speed in the horizontal direction), and the initial body vertical position speed (the initial upper body speed in the vertical direction).
- the calculation of this initial state is performed exploratively according to the flowchart of FIG. In the flowchart of FIG. 23, first, at S 200, based on the gait parameters of the normal gait (the parameters set at S 0 22 in FIG. 13), the desired foot position / posture, the desired arm
- the initial state (state at initial time Ts) of the posture and the desired body posture angle (tilt angle and angle) is determined, Kasumiga 09472
- the state represents the position or attitude angle and its rate of change (time derivative).
- the initial state of the target foot position / posture on the support leg side is based on the initial support leg foot posture of the first turning gait of the foot trajectory parameters determined in S100 in FIG.
- the foot position / posture trajectory (trajectory viewed from the next time's gait support leg coordinate system) leading to the second swing gait end free leg foot position / posture is determined by generating a finite time setting filter until time Ts.
- the initial state of the target foot posture of the swing leg is from the initial support leg foot position and posture of the current gait viewed from the next time's gait support leg coordinate system to the first swing gait end free leg foot position and posture.
- the trajectory of the foot position and posture is determined by generating a finite time settling filter until time Ts.
- the initial state of the target arm posture is determined to be the reference arm posture at time Ts obtained based on the reference arm posture trajectory parameters determined in S104 of FIG. More specifically, the position of the center of gravity of the both arms 5, 5 in the target arm posture (relative position with respect to the upper body 3), the distance between the left and right hands (tips of both arms 5 : 5) in the left and right direction and the reverse Phase swing angle and angular velocity are determined. However, the antiphase arm swing angle and angular velocity are corrected so that they continue at the boundary of the gait when the normal gait is repeated, as described later, so here they are only temporarily determined.
- the initial state of the target body posture angle is such that the reference body posture (tilt angle and angle) and the angular velocity at time Ts determined by the reference body posture trajectory parameter determined in S102 of FIG.
- the initial state of the desired body posture angle is determined.
- the initial state of the inclination angle (the inclination angle and its angular velocity) in the target body posture is zero.
- the desired foot position / posture trajectory, floor reaction force vertical component trajectory, and target ZMP trajectory of the normal gait are respectively the foot trajectory parameters and floor reaction determined in the flowchart in FIG. Force vertical component trajectory parameters, ZM It is determined independently of each other by the P orbital parameters.
- the instantaneous desired foot position / posture of the normal gait is determined according to the foot trajectory parameters without depending on the instantaneous value of the floor reaction force vertical component.
- the initial body horizontal position / velocity candidate ie, body horizontal position / velocity candidate at initial time Ts
- Xs, Vxs body horizontal position / velocity candidate at initial time Ts
- Xs, Vxs horizontal position, Vxs: horizontal velocity
- the provisional determination (Xs, Vxs) may be arbitrarily determined.
- the body horizontal position / velocity in the initial state of the normal gait obtained at the time of the previous generation of the gait is provisionally determined as candidates (Xs, Vxs). do it.
- the example of searching for the initial state of a normal gait in the X direction (forward and backward) on the sagittal plane is taken as an example. It is necessary to search for the initial state of the normal gait (the initial state that satisfies the boundary conditions of the normal gait) separately in the Y direction (horizontal direction) and the Y direction (left and right directions).
- a pseudo Jacobian sensitivity matrix
- the next candidate may be determined by the steepest descent method, or a simplex method may be used.
- the steepest descent method is used.
- a normal turning gait is temporarily generated (this temporarily generated normal turning gait may be hereinafter referred to as a temporary gait). More specifically, based on the gait parameters of the normal gait determined in S 0 22 in FIG. 13, the desired ZMP and the desired floor reaction force vertical at each moment from the initial time Ts to the end time Te The component, target foot position / posture, reference body posture, target arm posture, floor reaction force horizontal component allowable range, and floor reaction force moment vertical component allowable range are sequentially obtained. Then, the dynamic model (the model in Fig. 12) is used to satisfy the dynamic equilibrium condition between the obtained target ZMP and the target floor reaction force vertical component, and the condition of the floor reaction force horizontal component allowable range.
- the body position / posture is determined sequentially with the body horizontal position velocity (Xs, Vxs) and the body vertical position velocity (Zs, Vzs) as the initial state (time Ts) of the body 3. Generates a gait from time Ts to end time Te. At this time, the body posture is generated so as to match the reference body posture as much as possible.
- the anti-phase arm swing motion is determined so as to satisfy the condition regarding the floor reaction force moment vertical component, that is, the floor reaction camoment vertical component allowable range.
- the gait generation of the normal gait is only performed inside the gait generator 100, and is used as a target value for actually driving the robot 1 as a composite compliance operation determination unit 1 described later. It is never output to 04.
- the details of the normal gait generation processing by the sequential operation which is the processing of S208 will be described.
- FIG. 24 is a subroutine flowchart showing the process.
- the initial time Ts is substituted for the provisional gait generation time k.
- the body horizontal position velocity is currently provisionally determined (Xs, Vxs) (S2 in Fig. 23). 0 or the value determined in S 2 16 or S 2 18 described later), and the latest (Zs, V zs) obtained in S 206 is substituted for the body vertical position velocity. Is done.
- the reference body posture angle initial value (the angle at the initial time Ts) is substituted for the body posture angle
- the reference body posture angular velocity is the reference body body posture angular velocity initial value (the initial time Ts). Angular velocity) is substituted.
- the reference initial antiphase arm swing angle (angle at initial time Ts) is substituted for the antiphase arm swing angle
- the reference initial antiphase arm swing angular velocity angular velocity at initial time Ts
- S304 it is determined whether or not the provisional gait generation time k is before the gait end time (whether or not k ⁇ Ts + Tcyc). If the result of the determination is YES, the process proceeds to a subroutine for determining the instantaneous gait of S306, and the instantaneous gait value is determined. Next, the processing of the gait generator 100 proceeds to S308, and after increasing the provisional gait generation time k only, returns to S304.
- the value (current value) of the target ZMP trajectory at time k shown in FIG. 22 is obtained based on the normal gait parameters (ZMP trajectory parameters). Then, proceed to S404, based on the normal gait parameters (parameters for foot trajectory parameters, parametric parameters for upper body posture trajectories, and parameter parameters for arm posture trajectories), and the target both foot positions at time k.
- the posture (target foot position and posture on both the support leg side and the free leg side), the values of the reference body posture and the reference arm posture (current values) are obtained.
- the position of the overall center of gravity of the both arms 5, 5 (relative position with respect to the upper body 3), the distance between the left and right hands (tips of the both arms 5, 5)
- the value of the antiphase arm swing angle (this value) is obtained.
- the current value (the value at time k) of the target foot position / posture is obtained in the same manner as in the case where the foot position / posture at the initial time Ts is obtained in S200 in FIG.
- the value of the vertical position velocity of the center of gravity (this value) is calculated.
- the total body weight center-of-gravity vertical position velocity is calculated based on Expressions 01 and 04 relating to the dynamic model of FIG.
- Equations 01 and 04 the sum of the vertical center of gravity and the gravitational acceleration due to the motion of robot 1 multiplied by the total mass of robot 1 gives the floor reaction force vertical component.
- the relational expression of equality (the equation of motion in the vertical direction of the overall center of gravity of robot 1) is obtained. Therefore, from this relational expression and the desired floor reaction vertical component, the vertical acceleration of all body weights is obtained.
- the process proceeds to S408, and the body vertical position that satisfies the vertical position of the total body weight center is calculated.
- the body vertical position is calculated using Expression 04 relating to the model of FIG.
- the vertical position of the support leg material point 2 m and the free leg material point 2 m of the model in Fig. 12 is obtained from the current values of the target foot position and orientation on the support leg side and the free leg side.
- a vertical position of 3 m is required.
- the body vertical position of the body mass point of 3 m and the current value of the target body posture angle (the reference body posture angle set in S404 or the previous value determined in S414 described later ( From the target body posture angle at time k—A k), the body vertical position is obtained. 4 009472
- the current value of the desired body horizontal acceleration and the desired body posture acceleration is determined.
- the body horizontal acceleration and the body posture angular acceleration are determined so that the floor reaction force horizontal component Fx does not exceed [Fxmin, Fxmax].
- the current value of the target antiphase arm swing angular acceleration is determined so that the reaction force vertical component Mz does not exceed [Mzmin, Mzmax].
- one corner is determined to coincide with one of the reference body posture angles.
- components other than the anti-phase arm swing angle are determined so as to match the reference arm posture. At this time, while satisfying the above conditions, the target body tilt angle and the target anti-phase arm swing angle should follow as closely as possible the reference body tilt angle and the reference anti-phase arm swing angle, respectively. It is determined. The details will be described below.
- the instantaneous values (current values) of the foot position and posture and the body vertical position have been determined as described above.
- the arm posture components other than the antiphase arm swing angle are the reference arm posture. Is determined to match that of, therefore, the remaining body horizontal position, body posture angle and antiphase arm swing angle are determined 9 2, the target motion of the robot 1 can be uniquely determined. Therefore, all floor reaction forces are also uniquely determined.
- the desired floor reaction force vertical component and the desired ZMP of the normal gait are respectively the floor reaction force vertical component trajectory parameter and the desired ZMP trajectory parameter determined in S 0 22 in FIG. Specified by evening.
- the target ZMP is satisfied mainly by using the upper body tilt mode without using the upper body translation mode much (the horizontal component of the floor reaction force moment around the target ZMP is reduced to 0). ), The upper body posture angle may swing too much. Therefore, in order to prevent this, the body translation mode should be used as much as possible.
- the body translation mode involves changes in the floor reaction force horizontal component, if the floor reaction force horizontal component allowable range is narrow, there is a risk of slipping if the body translation mode is strongly activated. In this case, you have to rely on the upper body tilt mode.
- a gait that generates a floor reaction force horizontal component cannot be generated. I have to rely on Ido.
- the anti-phase arm swing motion can change only the floor reaction force moment vertical component without changing both the horizontal component of the floor reaction force moment around the target ZMP and the floor reaction force horizontal component. It can be used to prevent the reaction force vertical component from exceeding the allowable range of the floor reaction vertical component.
- the body horizontal acceleration, the body posture angle acceleration, and the antiphase arm swing acceleration are determined according to the flowchart shown in FIG.
- the determination of the body horizontal acceleration and the body posture angular acceleration angular acceleration of the inclination angle of the body 3 was performed in the X direction (front-back direction) on the sagittal plane.
- the body horizontal acceleration and body posture angular acceleration in the Y direction are also determined in the same manner as in the X direction.
- the value of the reference body at time k is substituted for the target body. Also, the value of the reference arm posture at time k is substituted for the target arm posture, excluding the antiphase arm swing angle and angular velocity components of the arm posture.
- the current time (the value of the timer for creating a normal gait) k is the body posture angle and the antiphase arm swing angle restoration period (in the case of a normal gait, the body posture angle and antiphase arm swing) It is determined whether the angular restoration period is between the time Tm and the time Ts2 and the time from the time Tm2 to Te.). If the determination result of S 5 02 is N ⁇ , the process proceeds to S 5 0 4, and if it is Y E S, the process proceeds to S 5 30.
- the vertical position of the supporting leg material point 2 m and the free leg material point 2 m is obtained using the target foot position and orientation at the current time k (this time).
- the vertical position of the upper body mass point 3 m is obtained using the floor reaction force vertical position at the current time k (this time), and the time series value of the target body vertical position obtained up to the current time k is used.
- the vertical acceleration of the upper body mass point 3 m at the current time k is obtained. Then, by substituting these calculated values into the above equation 0 3 y and solving the equation of My equation 0 3 y in which d2 0 by / dt2 is 0 for d2Xb / dt2, the upper body mass point is obtained.
- Horizontal acceleration is upper body 2004/009472
- the reference body posture trajectory parameter is set so that the inclination angle of the reference body posture changes
- the reference body posture angular acceleration at the current time k (the body 3 If the reference angular acceleration of the tilt angle is not 0, the angular acceleration in the body tilt mode is set to the value of the non-zero reference body posture angular acceleration, and the body horizontal acceleration a tmp is obtained using a dynamic model.
- d2 0 by / dt2 in Equation 03 y is set to a non-zero reference body posture angular acceleration, and the body horizontal acceleration a tmp is obtained in the same manner as described above.
- the floor reaction force horizontal component Fxtmp at time k when the body horizontal acceleration is tmp is obtained using a dynamic model.
- Fxtmp is obtained using the equation 02X of the dynamic model. That is, Fxtmp is obtained by the following equation (17).
- d2Xsup / dt2 and d2XswgZ dt2 represent the horizontal acceleration of the support leg foot mass point and the free leg foot mass point at time k, respectively.
- Figure 27 shows an example of Fxtmp obtained in this way.
- Figure 27 shows the part where Fxtmp exceeds the floor reaction force horizontal component allowable range [Fxmin, Fxmax]. 004/009472
- the body horizontal acceleration a in the body translation mode the floor reaction force horizontal component Fx generated by this, the body angle acceleration in the body tilt mode / 3 are determined as follows: (S508 to S516).
- Fx Fxmin Equation 1 9 Otherwise, that is, if Fxtmp is within the floor reaction force horizontal component allowable range [Fxmin, Fxmax], the process proceeds to S5 14 and Fx is determined by the following equation.
- Fx Fxtmp ... Equation 20
- body posture angular acceleration body inclination angular acceleration 3
- K a tmp + (F-Fxtmp) / ⁇ Fp ... Equation 2 1
- the body translation mode and the body body acceleration mode ⁇ obtained above are calculated. It is better to determine the upper body horizontal acceleration ⁇ in the upper body translation mode analytically or exploratively using a stricter dynamic model so that the synthesized motion satisfies the target ZMP.
- a pseudo Jacobian sensitivity matrix
- the next candidate may be determined by a pseudo Newton method or a simplex method.
- Fx Fxmax and the floor reaction force around the target ZMP.
- the set of acceleration / 3 may be searched for.
- Figure 28 shows the Fx obtained as described above.
- FIG. 29 shows the body posture angular acceleration obtained as described above. P hire 004 ⁇ 72
- the upper body acceleration (upper body angular acceleration) is the reference unit angular acceleration d20 bzref / dt2
- antiphase arm swing angular acceleration) 3a is the reference antiphase arm swing angular acceleration d2 ⁇ azref / dt2
- the floor reaction force moment vertical component Mztmp is calculated.
- d20 bzref / dt2 is j3 bref and d2 ⁇ azref / dt2 is / 3 aref.
- Mztmp obtained by substituting Expression 1001 into Expression 1004 into Expression 03z is Mztmp.
- Xb and Yb are the horizontal body at time k-1 The position is substituted, and the value of the time k is substituted for Xzmp, Yzmp, Xsup, d2Ysup / dt2, Xswg, and d2Yswg / dt2.
- Figure 32 shows an example of Mztmp obtained in this way.
- the part where Mztmp exceeds the permissible range of the floor reaction force moment vertical component [Mzmin, Mmax] is indicated by oblique lines.
- Mztmp> Mzmax the process proceeds to S522, and Mz is determined by the following equation.
- Mztmp ⁇ Mzmin proceed to S524, and Mz is determined by the following equation.
- Mz Mzmin ... Equation 1 0 1 9
- Mztmp Mzmin ... Equation 1 0 1 9
- Mz obtained as described above indicates a vertical component of floor anti-camo due to the motion of the entire rod including the antiphase arm swing.
- the antiphase arm swing angular acceleration ⁇ a is determined so that this Mz does not exceed the floor reaction force moment vertical component allowable range [Mzmin, Mzmax].
- the anti-phase arm swing angular acceleration ⁇ a is calculated by dividing Maz by the equivalent inertia moment A of anti-phase arm swing A Maz. Value). That is,] 3a is obtained by the above expression 1021.
- Figure 35 shows the antiphase arm swing angular acceleration] 3a.
- the vertical component Mz of the floor reaction force moment generated by the motion of the entire rod including the antiphase arm swing is within the allowable range [Mzmin, Mzmax].
- the anti-phase arm swing angle acceleration is equal to the reference anti-phase arm swing angular acceleration ⁇ aref.
- the floor reaction force moment vertical component Mztmp cancels out of the allowable range.
- antiphase arm swing angular acceleration] 3 a is determined.
- floor reaction force moment vertical component Mz strictly exceeds floor reaction camoment vertical component allowable range [Mzmin, Mzmax].
- antiphase arm swing angular acceleration) 3a is determined analytically or exploratively using a stricter dynamic model. Better.
- a pseudo Jacobian sensitivity matrix
- the next candidate may be determined by the pseudo Newton method or the simplex method.
- the above is the processing when the time k is not during the body posture angle / opposite phase arm swing angle restoration period.
- the determination result of S502 is YESS
- the following processing is performed. First, proceeding to S530, from the previous instantaneous gait state of robot 1 (gait state at time k-1), the angular acceleration in the body tilt mode is set to 0, and the motion in the body translation mode is performed.
- the mouth port 1 is set, the body horizontal acceleration ⁇ required to satisfy the target ZMP at this time (time k) is obtained, and this is determined as the final body horizontal acceleration.
- the process proceeds to S532, in which case the floor reaction force horizontal component Fx is obtained.
- the flow proceeds to S534, and the body posture angular acceleration (body inclination angular acceleration) ⁇ is determined to be zero.
- the upper body angular acceleration is determined to be the reference upper body angular acceleration) 3 bref (the second derivative of the reference upper body angular).
- Substitute out-of-phase arm swing angular acceleration j8 are f (the second-order derivative of the reference out-of-phase arm swing angle).
- the body posture angular acceleration (body inclination angle acceleration and body tilt angle acceleration) is used as the reference body posture angular acceleration
- the antiphase arm swing angular acceleration is used as the reference antiphase arm swing. Match the angular acceleration. Even in this case, it is expected that the floor reaction force generated by the movement will not exceed both the floor reaction force horizontal component allowable range and the floor reaction force moment vertical component allowable range. no problem.
- the anti-phase arm swing velocity is obtained by successively integrating the anti-phase arm swing acceleration J 3 a obtained in S 4 12 (cumulative addition from time Ts to the current time k). Then, this is sequentially integrated (cumulative addition from time Ts to current time k) to obtain the antiphase arm swing angle S az (this time value).
- the ZMP-converted value of the floor reaction force moment that generates the body posture angular acceleration to return the body posture angular velocity to the initial value (the value at time Ts) by time Te (Hereinafter referred to as ZMP converted value of body posture restoring moment and abbreviated as ZMPrec).
- Body posture angle ⁇ In the antiphase arm swing angle restoration period (the period from time Tm to time Ts2 and the period from time Tm2 to Te), the body posture angular acceleration is generated using the body tilt mode.
- the ZMP (k) calculated from the motion is shifted by ⁇ ZMP obtained by the following equation. .
- AZ MP (k) -) 3 (k) * ⁇ ⁇ / Fz (k)... Equation 2 3 Therefore, if the pattern of ⁇ r and the pattern of Fz (k) are determined (if known), ⁇ By setting the ZMP (k) pattern appropriately Then, a body posture angular acceleration pattern that satisfies Equation 23 is generated, and the body posture angular velocity is set to the initial value (value at time Ts), that is, the body in the initial (time Ts) state of the reference body posture trajectory. It can return to the attitude angular velocity.
- the body posture restoring moment ZMP converted value (ZMPrec) means ⁇ ZMP (k) appropriately set as such.
- AMr fluctuates strictly, but may be approximately a constant value. This is because the normal gait is only generated temporarily and does not cause the actual mouth pot to follow this gait, so the dynamic accuracy of the normal gait does not need to be very high.
- Figure 30 shows an example of Z MPrec.
- a trapezoid pattern is formed as a ZMPrec pattern in a period from time Tm to time Ts2 and a period from time Tm2 to Te.
- the time at the break point of the trapezoid is matched with the break point time of the target ZMP pattern in the period between time Tm and time Ts2 and in the period from time Tm2 to Te (see Fig. 22). This is because, as will be described later, it is easy to correct the target ZMP pattern of the gait this time.
- Equation 24 the / 3 (k) obtained by Equation 24 is as shown by the solid line in FIG.
- the dotted line in Fig. 31 shows the body posture angular acceleration during the period from time Ts to time Tm and the period from time Tm2 to Te (shown by the solid line in Fig. 29). (If it is clear that the value is at time k, (k) may be omitted.)
- the initial (time Ts) body posture angle is equal to the initial (time Ts). Reference body posture angle.
- the initial body posture angular velocity is determined so as to satisfy Equations 37a and 37b. End body posture angle-Initial body posture angle
- Equation 3 7a Terminal body posture angular velocity-Initial body posture angular velocity
- Equation 3 7b The integration period of the first term on the right side of each of Equations 37 a and 37 b is the sum of the period from time Ts to Tm and the period from Ts2 to Tm2.
- the integration period of the second term on the right side is the sum of the period from time Tm to Ts2 and the period from Tm2 to Te.
- the initial state posture angle and the angular velocity viewed from the support leg coordinate system (the next time's gait supporter. Coordinate system) of the first turning gait are the following first gait, respectively. It must be the same as the terminal body posture angle and angular velocity seen from the supporting gait coordinate system of the turning gait (the next and next gait supporting leg coordinate system). Therefore, in this reference example, the initial (time Ts) body posture angle is determined to be the value of the initial (time Ts) reference body posture angle, and this is used as the value of mouth port 1 in the normal gait.
- Equation 37a The left-hand side of Equation 37a is the result of the coordinate transformation to the value seen from the next time's gait support leg coordinate system using a matrix (rotation coordinate transformation matrix) corresponding to the rotation angle of the barrel (the rotation angle around the vertical axis). Substitute the initial body posture angle and the end body posture angle of. Further, the body posture angular acceleration relating to the integration of the first term on the right side of each of the expressions 37a and 37b is the one obtained in step S518 of FIG.
- the initial body posture angular velocity of Equations 37a and 37b, and the trapezoid height of ZM Prec (the trapezoidal pattern in FIG. 30) relating to the integration of the second term on the right side of Equations 37a and 37b
- the time of the break point of the trapezoidal pattern of Z MPrec is determined in advance as described above.
- the trapezoid height acycl of ZM Prec of the first turning gait and the ZMP rec of the second turning gait are The trapezoid height acyc2 shall be the same value.
- the initial body posture angular velocity obtained by solving the simultaneous equations of equations 37a and 37b including those unknowns will be used as the new initial body posture angular velocity. decide.
- the final body posture angular velocity in Equation 37b is obtained by coordinating the unknown initial body posture angular velocity to the value seen from the next support leg coordinate system using a matrix corresponding to the total turning angle of the normal gait. It has been converted.
- the body posture angle acceleration ⁇ becomes as follows.
- ] 3 — ZMP rec * FzZ ⁇ r ⁇ ⁇ ⁇ Equation 1 0 2 5
- the body horizontal acceleration that satisfies the target ZMP when the body tilt restoring moment is not generated is QUmp as obtained in S5 32,
- the body horizontal acceleration required to satisfy the target ZMP is as follows.
- the acceleration increases by the second term on the right-hand side of Equation 1027 due to the tilt restoration moment ⁇ ⁇ ⁇ converted value (ZMP rec).
- the above-mentioned extreme body tilt restoration moment ZMP converted value (ZM Prec) The end body when the body posture angle acceleration j8 is changed so as to generate a pattern
- the horizontal speed is the terminal upper body horizontal speed when the body tilt restoration moment ZMP conversion value (ZM Prec) pattern is not generated, that is, the terminal value of the upper body horizontal speed obtained in S4 14 It is obtained by adding the first-order integration of (ZM Prec * FzZ AMp) up to the time Ts force and Te.
- the body tilt restoration moment When the body posture angle acceleration] 3 is changed to generate the ZMP converted value (ZM Prec) pattern
- the end body horizontal position is changed to the ZMP converted value (ZM Prec) pattern.
- the second-order integral of (ZM Prec * FzZ ⁇ ) from time Ts to Te is added to the terminal body horizontal position when there is no Desired.
- the process proceeds to S314, and the antiphase arm swing restoration angular acceleration (/ 3 arec) pattern is determined so that the antiphase arm swing angular velocity coincides with the initial and end positions.
- the antiphase arm swing restoration angular acceleration pattern is set to a trapezoidal shape as shown in Fig. 36, and the trapezoid height azcyc2 during the period from time Tm to Ts2 and the trapezoidal height during the period from time Tm2 to Te azcycl is the same, and from the time Ts to Te; the integral of 8 arec and the antiphase arm swing acceleration
- 3a are calculated so that the vertical component of floor anti-camoment Mz does not exceed the allowable range.
- the floor reaction force moment vertical component (Mazrec) generated by the antiphase arm swing restoration angular acceleration pattern determined in this way is as shown in Fig. 37. Therefore, the vertical component Mz of the floor reaction force moment generated by the movement of the mouth pot including the antiphase arm swing finally becomes Mztmp in Fig. 32 and Maz in Fig. 34 as shown in Fig. 38.
- the sum of 37 and Mazrec that is, the sum of Mz in Figure 33 and Mazrec in Figure 37.
- trapezoidal restoration moments are added, but these periods are set to have a sufficiently wide allowable range. Therefore, the vertical component of the floor reaction force moment generated by the movement of the mouth pot including the antiphase arm swing must not exceed the allowable range. And not.
- Initial antiphase arm swing angular velocity is determined by the following equation.
- the antiphase arm swing angle when 3 arec is 0 is the antiphase arm swing angle (antiphase arm swing angle at time Te) obtained in S 16. That is.
- the second-order integral of 13 arec is the second-order integral of the antiphase arm swing restoration angular acceleration from time Ts to Te set as shown in Fig. 36.
- the reference initial antiphase arm swing angular velocity is a value at the time Ts of the reference antiphase arm swing angular velocity (first-order differential value of the reference antiphase arm swing angle 0 aref).
- the initial anti-phase arm swing angle may be made equal to the reference initial anti-phase arm swing angle, or the anti-phase arm swing angular acceleration finally determined (that is, the floor reaction force, the moment vertical component Mz is allowable) Based on the calculated antiphase arm swing acceleration) 3a and the restored angular acceleration / 3 arec) to prevent the arm from exceeding the range, and the obtained initial antiphase arm swing angular velocity, the initial antiphase arm
- the average value of the difference between the arm swing angle calculated when the swing angle matches the reference initial antiphase arm swing angle and the average value of the maximum value and the minimum value of the difference is calculated.
- a value obtained by subtracting one half of the obtained average value from the reference initial antiphase arm swing angle may be determined as the final initial antiphase arm swing angle.
- the body posture angular velocity in the period from the time Tm to Ts2 and the period from the time Tm2 to Te is set to the initial body posture trajectory. This is to prevent the floor reaction force horizontal component Fx from exceeding the allowable range [Fxmin, Fxmax] even when the body posture angular acceleration 3 is generated so as to return to the angular velocity.
- the allowable range of the floor reaction force horizontal component is sufficiently large, and the body posture angular velocity should be returned while satisfying the target ZMP. Even when the body posture angular acceleration j3 is generated, the floor reaction horizontal component F x does not exceed the allowable range.
- Another reason for setting the times Ts, Tm, Ts2, and Tm2 as described above is that, in the period from the time Tm to Ts2 and the period from the time Tm2 to Te, the antiphase arm swing angular velocity is set to the reference antiphase arm. This is to prevent the floor reaction moment vertical component Mz from exceeding the allowable range [Mzmin, Mzmax] even if the antiphase arm swing angular acceleration a is generated so as to return to the initial angular velocity of the swing orbit. .
- the floor reaction force moment vertical component allowable range is sufficiently large, so that the anti-phase arm swing is returned to return the anti-phase arm swing angular velocity. Even if 3a is generated, the floor reaction force moment vertical component Mz does not exceed the allowable range.
- the process proceeds to S210 in FIG. 23, and the terminal body horizontal position and speed of the generated gait (assumed normal gait) are It is converted to the value viewed from the support leg coordinate system ( ⁇ '", ⁇ '", ⁇ '"coordinate system in Fig. 17) corresponding to the instantaneous support leg, and the value is set as (X e, VX e) (X e: end body horizontal position, V xe: end body horizontal velocity). Then, the process proceeds to S212, where the difference between the initial body horizontal position speed (Xs, VXs) and the terminal body horizontal position speed (Xe, VXe) is calculated as shown.
- the flow proceeds to S2114, and it is determined whether or not the calculated body horizontal position / velocity boundary condition error (errx, errvx) is within an allowable range set in advance as appropriate.
- the initial divergent component (Xs + VXs0> 0) and the terminal divergent component (Xe + VxeZoO)
- the terminal divergent component (Xe + VxeZoO)
- ⁇ ' is a predetermined value as described above.
- S2 16 a plurality (two in this reference example) of initial value candidates (X s + AX s, VX s), (X s, V xs' + ⁇ ) are provided near (X s, VX s) V xs) is determined.
- AX s and AV xs mean predetermined small changes with respect to X s and V xs, respectively. Then, using each of these initial value candidates as the initial state of the body horizontal position / velocity, a normal gait is generated using the gait parameters by the same processing as in S208.
- the end body position velocity of the generated normal gait is calculated from the supporting leg coordinate system ( ⁇ '", ⁇ '", ⁇ '"coordinate system in Fig. 17) corresponding to the supporting leg at that moment. Convert to the value you saw (Xe + AXel, Vxe + AVxe1) and (Xe + room Xe2, Vxe + ⁇ Vxe2) are obtained.
- (X e + AX el, V xe + ⁇ V xe 1) means the end body position velocity corresponding to (X s + AX s, V xs) (, and (X e + AX e 2, V xe + A Vx e 2) means the terminal body position velocity corresponding to (X s, V xs + ⁇ V xs).
- the initial state of the variables other than the body horizontal position velocity is, for example, the initial value of the body horizontal position velocity candidate. May be set the same as when (X s, Vx s).
- each initial value candidate Xs + ⁇ Xs, Vxs
- each initial value candidate Xs + AVxs
- the body horizontal position / velocity boundary condition error corresponding to each of them is obtained.
- the process returns to S206.
- the above processing (the processing of S206 to S218) is repeated as long as the determination result of S214 is NO.
- S300 in the process (S208) of generating a normal gait corresponding to the new initial value candidate (Xs, Vxs) of the body horizontal position / velocity,
- the initial value of the body posture angular velocity is not the initial value of the reference body posture angular velocity, but the S3 in the processing of S208 corresponding to the previous initial value candidate (Xs, Vxs) of the body horizontal position velocity. It is set to the value determined in 10 (see Fig.
- (X0, V0) and (Z0, Vz0) are the body leaning restoring moment ZMP converted value pattern determined in S310, and the normal gait at time Ts. Based on the initial body posture angle and angular velocity, and the body horizontal position velocity (Xs, Vxs) at time Ts after the loop of S204 was removed, the gait was adjusted to satisfy the gait conditions.
- the body vertical position and the velocity are defined by the supporting leg coordinate system corresponding to the supporting leg of one step starting from time Tcyc (that is, the first turning gait for the second time) ( ⁇ '", ⁇ '", ⁇ "in Fig. 17)
- the body posture angle and angular acceleration measured are calculated using the support leg coordinate system (X "', ⁇ '", ⁇ ) corresponding to the support leg of one step starting from time Tcyc (that is, the first turning gait for the second time).
- '"Coordinate system is determined as the value converted to the value.
- ⁇ is a certain value as described in the explanation about divergence.
- the initial antiphase arm swing angle and the angular velocity (0 azO, coazO) at the original initial time 0 are obtained.
- the value (0 a Z O ", ⁇ az O") which is a value viewed from the supporting leg coordinate system is obtained.
- (0 az0, O) az0) was determined in S314 and S316.
- Antiphase arm swing restoration angular acceleration pattern initial (time Ts) inverse of normal gait
- the anti-phase arm swing angle trajectory is determined so that the vertical component of floor anti-camoment does not exceed the allowable range, and the body posture angle and anti-phase arm swing angle are restored.
- the antiphase arm swing angle trajectory is determined so that the sum of the reference antiphase arm swing angular acceleration ⁇ aref and the antiphase arm swing restoration angular acceleration ⁇ arec is generated).
- time Te-Ts time Te-Ts
- the parameters of the foot trajectory of the current time gait are set so that the foot position / posture trajectory of the current time gait is connected to the foot position / posture trajectory of the normal gait.
- the initial swing leg foot position / posture of the current time gait (the initial value of the current time gait swing leg foot position / posture) is the current free leg position / posture (previous time gait viewed from the gait support leg coordinate system). (End position of free swing leg).
- the current gait support leg foot position / posture (the initial value of the current time gait support leg foot position / posture) is the current support leg foot position / posture viewed from the gait support leg coordinate system (previous gait end support leg) (Foot position / posture).
- the current gait end swing leg foot position / posture is calculated using the next time's gait support leg coordinate system viewed from the current time's gait support leg coordinate system (the required value of the first step related to the current time's gait. ). That is, now PC leak 004/009472
- the support leg foot 22 is off the floor and in the air.
- Support leg foot 22 In order to determine the trajectory after the foot 2 has left the floor, the support leg foot landing scheduled position and posture are set.
- the planned landing position / posture of the supporting leg foot corresponds to the coordinates of the next time's gait supporting leg (the required value of the free leg foot position / posture of the second step related to the current time's gait) as viewed from the coordinates of the gait supporting leg this time. Is set. More specifically, the support leg foot landing expected position / posture is determined based on the position / posture, and the foot 22 is placed almost entirely on the floor so that the foot 22 does not slip while keeping the foot 22 in contact with the floor. The representative point of the foot 22 when rotated until it touches the ground is set to match the origin of the next-time gait support leg coordinates viewed from the current time gait support leg coordinates.
- the current gait end support leg foot position / posture is calculated from the current support leg position / posture (current gait initial support leg foot position / posture) to the foot landing scheduled position / posture corresponding to the next time gait support leg coordinate system (as described above).
- the trajectory of the foot position / posture that leads to the parameter can be obtained by generating the finite time settling filter to the end of the gait this time.
- the flow proceeds to S602, and the reference body posture trajectory parameter of the current time's gait is determined in the same manner as the first turning gait / the second turning gait of the normal gait.
- the reference body posture trajectory of the current time's gait is continuously connected to the reference body posture trajectory of the normal gait.
- the reference body posture angle and angular velocity at the end of the current time's gait are respectively The above parameters are set to match the body posture angle and angular velocity.
- the reference body posture regarding the inclination angle is a steady vertical posture in both the current time gait and the normal gait. 00 value 9472
- the reference arm posture trajectory parameter of the current time's gait is determined in the same manner as the first turning gait / the second turning gait of the normal gait.
- the initial reference arm posture of this time's gait and its change rate match the current instantaneous values of the reference arm posture and change rate, and the arm posture trajectory of this time's gait is continuous with the arm posture trajectory of the normal gait.
- the arm posture trajectory parameters determined here are the same as those for the determination of the normal gait parameters (S104 in Fig. 15).
- the center of gravity of both arms 5,5 The parameters (position relative to the upper body 3), the distance between the left and right hands (the ends of both arms 5, 5) in the left and right direction, and the parameters related to the antiphase arm swing angle are determined.
- the position of the center of gravity of the entire arms 5 is set so as to be kept constant with respect to the upper body 3.
- the floor reaction force vertical component trajectory parameters of the gait this time are defined by the parameters, as in the case of the first gait ⁇ the second gait of the normal gait.
- the floor reaction force vertical component trajectory is set so as to be a substantially continuous trajectory (values do not fly stepwise) as shown in FIG.
- the floor reaction force vertical component trajectory parameters are determined so that both the vertical position velocity of the overall center of gravity of the gait and the floor reaction force vertical component trajectory are continuously connected to the normal gait.
- the initial body vertical position / velocity of the normal gait finally obtained in the processing of S 0 24 in FIG.
- the values (Z0 ", Vz0") converted into the values viewed from the coordinate system that is, (Z0 ", Vz0") obtained in S224 of Fig. 23, etc.
- the initial overall center-of-gravity vertical position velocity of the normal gait viewed from the gait support leg coordinate system is obtained using Equation 04 (or the kinematics model of Robot 1) described above, for example.
- the vertical position of the initial total body weight of the normal gait of the normal gait viewed from the gait support leg coordinate system is the upper body vertical position Z0 "of the normal gait obtained in S224.
- the vertical position of the upper body mass point of the model in Fig. 12 corresponding to Fig. 12 and the positions of the respective feet at the beginning of the normal gait are converted to the values viewed from the gait support leg coordinate system this time. It can be obtained by substituting the leg body mass point vertical position on the leg side into Equation 04.
- the vertical velocity of the initial overall center of gravity of the normal gait viewed from the gait support leg coordinate system in this case is the upper body vertical velocity VzO "of the normal gait obtained in S224, as shown in the model of Fig. 12.
- Equations 41a and 41b are the integral values during the period from the beginning to the end of the current time's gait.
- At least two of the parameters of the floor reaction force vertical component parameters (such as breakpoint times) that define the floor reaction force vertical component pattern as shown in Fig. 6 are independent.
- the value of the unknown variable is determined by solving a simultaneous equation composed of equations 41a and 41b.
- the height (peak value of the floor reaction force vertical component) and width (single leg support period) of the trapezoid in Fig. 6 may be selected as the parameters of the floor reaction force vertical component as unknown variables.
- the inclination of both sides of the trapezoid in Fig. 6 should be a value determined in advance according to the gait cycle this time, or the floor reaction force vertical component pattern excluding the time of transition from the one-leg support period to the air period.
- the time of the point is a value determined in advance according to the current time's gait cycle and the like. Supplementally, assuming one unknown variable, there is generally no solution that satisfies the simultaneous equations of Equations 41a and 41b.
- the routine proceeds to S608, where the floor reaction force horizontal component allowable range [Fxmin, Fxmax] (specifically, a parameter defining the pattern of the floor reaction force horizontal component allowable range) is the 1 turning gait / 2nd turning gait.
- the floor reaction force horizontal component allowable range is set in a pattern as shown in FIG.
- the floor reaction force horizontal component and the permissible range are set based on the above equation 12 according to the floor reaction force vertical component pattern previously determined in S606.
- the floor reaction force moment vertical component allowable range [Mzmin, Mzmax] (specifically, the parameter that defines the pattern of the floor reaction force moment vertical component allowable range) is the first turning gait of the normal gait. ⁇ ⁇ ⁇ It is set in the same way as the second turning gait. For example, in the pattern shown in Figure 41 2004/009472
- the floor reaction force moment vertical component allowable range is set.
- the floor reaction force moment vertical component allowable range is set based on the above-mentioned equation 1102 according to the floor reaction force vertical component pattern determined in S606 previously.
- the ZMP trajectory of this time's gait (specifically, the parameters that define the ZMP trajectory, the time and position of the break point of the gait) are Turning gait—Similar to the second turning gait, the gait is set as shown in FIG. 7 so that the stability margin is high and there is no sudden change.
- the above parameters are set so that the ZMP trajectory of the current time gait is continuously connected to the ZMP trajectory of the normal gait. That is, the ZMP trajectory parameters are determined so that the ZMP position at the end of the gait this time matches the ZMP position at the beginning of the normal gait.
- the method of setting the time and position of the break point of the ZMP trajectory during the one-leg support period may be the same as the method of setting the ZMP trajectory parameters of the normal gait described above. Then, the ZMP trajectory parameters should be set so that the target ZMP trajectory in the aerial period changes linearly and continuously from the start of the aerial period to the ZMP position at the beginning of the normal gait.
- the parameters of the ZMP trajectory of the current time's gait determined in S612 are only provisionally determined and will be corrected as described later. Therefore, the ZMP trajectory of the current time's gait set above will be referred to as the temporary target ZMP trajectory of the current time's gait.
- the body posture angle and antiphase arm swing angle restoration period [Ta, Tb] are set.
- the body posture angle and antiphase arm swing angle restoration start time Ta is equivalent to Tm in the second turning gait of the normal gait
- the body posture angle and antiphase arm swing angle restoration completion time Tb is the normal gait This is equivalent to Ts2 in the second turning gait.
- the method of setting these times Ta and Tb is the same as the method of setting Tm and Ts2, respectively.
- ZMP orbit parameters ZMP orbit parameters
- ZMP trajectory parameters are corrected to make the upper body posture trajectory continuous or close to the normal gait, and the anti-phase arm swing angle converges to the anti-phase arm swing angle trajectory of the normal gait In this way, the parameters related to the antiphase arm swing angle of the gait this time are determined.
- FIG. 42 is a subroutine flow chart showing the processing. First, the process proceeds to S702 via S700, and a temporary current time gait up to the current time gait end time is temporarily generated based on the temporary target ZMP pattern and other current time gait parameters.
- the terminal state of the previous target gait (more specifically, the gait state such as the body horizontal position speed, the body vertical position speed, the body posture angle and its angular velocity, the target foot position / posture, the target arm posture, etc.) Is converted to the support leg coordinate system this time, and this is the initial state of the gait this time.
- the target arm posture includes the target antiphase arm swing angle and angular velocity.
- S804 it is determined whether or not the provisional gait generation time k is before the current time gait end time Tcurr (whether or not k ⁇ Tcurr). If the result is YES, the process proceeds to the current gait instantaneous value determination subroutine of S806, and the instantaneous value of the current time's gait at time k is determined. In the subroutine for determining the gait instantaneous value of S806, the aforementioned S306 and 2004/009472
- the process proceeds to S706, where the terminal divergence component error, which is the difference between the gait terminal divergence component q0 [k] and the normal gait initial divergence component Q "(determined in S224 of FIG. 23) errq is obtained using the equation shown: Further, the process proceeds to S708, and it is determined whether or not the obtained terminal divergence component error errq is within an allowable range (a range near 0).
- the terminal divergence component error which is the difference between the gait terminal divergence component q0 [k] and the normal gait initial divergence component Q "(determined in S224 of FIG. 23) errq is obtained using the equation shown: Further, the process proceeds to S708, and it is determined whether or not the obtained terminal divergence component error errq is within an allowable range (a range near 0).
- a Aa (Aa is a predetermined minute amount), and a trapezoidal shape is added to the current temporary target ZMP pattern according to the relationship in FIG. Based on the corrected target ZMP, a provisional current time gait up to the end is calculated in the same manner as in S702.
- a is to make the gait end divergence component of the current time gait coincide with the normal gait initial divergence component as much as possible. This is the height of the trapezoidal pan for correcting the tentative target ZMP, in order to approach the orbit.
- the correction of the provisional target ZMP is performed during the period in which almost the entire bottom surface of the support leg foot 22 is in contact with the ground (period of sole contact), that is, the allowable range of the floor reaction force horizontal component is sufficient.
- the time at the break point of the trapezoidal pattern is set in accordance with the time of the break point of the tentative target ZMP during the entire sole contact period.
- the terminal divergence component ql [k] in the provisional current time gait is calculated using the equation (formula 10 above) based on the body horizontal position velocity (X e 1, VX el) at the provisional current time gait end. Required.
- the value of 3 is a small constant that is appropriately set in this reference example.
- ⁇ a ⁇ a may be set so as to decrease .DELTA.a.
- the terminal divergence component error errq can be kept within an allowable range by several repetition operations. Then, the process proceeds to S 714, where the parameter sensitivity r (the rate of change of the terminal divergent component error with respect to ⁇ a) is obtained from the equation shown.
- a -errq / r, that is, the value obtained by dividing the terminal divergence component error errq obtained in S 706 by the parameter sensitivity r obtained in S 714 is a height a.
- the temporary target ZMP pattern is corrected by adding the correction amount of the trapezoidal pattern to the temporary target ZMP pattern according to the relationship shown in FIG. 44 (a new temporary target ZMP pattern is determined).
- the difference between the terminal body posture angle of the provisional current time gait and the initial body posture angle of the normal gait, and the terminal body posture angular velocity of the provisional current time gait and the initial body of the normal gait Based on the difference from the posture angular velocity, etc., the body posture restoration moment of this gait ZMP conversion value (ZMP) so that the body posture angle trajectory of the gait this time approaches the body posture angle trajectory of the normal gait MPrec) pattern is determined.
- ZMPrec determined here is used during the generation of the instantaneous gait instantaneous value, which will be described later, during the period in which the floor reaction water water allowable range is sufficiently large (the period during the one-leg support period).
- This ZMPrec is a trapezoidal pattern similar to that described in the normal gait generation processing, and is specifically determined as follows. That is, like the trapezoidal pattern of ZM Prec in the period of the second turning gait in Fig. 30, the trapezoidal pattern of ZMPrec of this time's gait is set, and the time of the apex (break point) of the trapezoid is assumed to be known. Specifically, the trapezoid break point time is adjusted to the target ZMP break point time), and the trapezoid height is set to an unknown value, and the trapezoid height (parameter) of ZMPrec is calculated as follows. However, the time at which the trapezoidal pattern of Z M Prec starts to rise is Ta, and the time at which the trapezoidal pattern returns to 0 is Tb.
- both the body posture angle and the body posture angular velocity are continuous to the normal gait at the end of the gait this time. Is generally not possible. Therefore, in this reference example, the unknown parameters are determined so that the state of the generated gait gradually approaches the state of the normal gait over a plurality of steps.
- PC orchid 004/009472
- the ZM Prec pattern in one gait is complicated, the number of unknown parameters is set to two or more, and both the body posture angle and the body posture angular velocity are calculated at the end of the gait.
- the gait may be connected continuously, but the ZMP rec pattern may fluctuate too much zigzag.
- the difference between the terminal body posture angle of the provisional current time's gait and the initial body posture angle of the normal gait obtained by setting the trapezoid height of the ZM Prec pattern to 0 in S702 is calculated as 0. err. Also, the difference between the terminal body posture angular velocity of the provisional current time gait and the initial body posture angular velocity of the normal gait is obtained, and this is set as v0 err.
- the current time's gait is generated with the trapezoid height of the ZM Prec pattern being a certain value bcurr, and that the first turning gait is subsequently generated by the same algorithm as the current time's gait.
- the body posture restoring moment of the first turning gait ZMP converted value ZM Prec is the ZMP rec pattern of the first turning gait obtained in S310 of FIG. It is assumed that the sum is the sum of the acycling (trapezoidal pattern in Fig. 30) and a certain value bl.
- the gait generated in this way is called a ZM Prec modified gait, and its end (the end of the first turning gait)
- the body posture angle and the angular velocity are 1, ⁇ ⁇ , respectively.
- the original normal gait obtained when the subroutine processing for obtaining the initial state of the normal gait in S024 is completed (the normal body gait initial body posture angle and the normal gait finally determined in S310)
- the upper body posture angle and the angular velocity are ⁇ lorg and ⁇ lorg, respectively.
- ⁇ 0 1 and ⁇ 1 are defined as follows. twenty five
- ⁇ 1 - ⁇ ⁇ 1- ⁇ ⁇ lorg... Equation 5 1
- ⁇ 0 1 and ⁇ 1 are the difference between the body posture angle and the angular velocity difference between the ⁇ M Pi'ec corrected gait and the original normal gait up to the end of the first turning gait. Means If ⁇ 0 1 and ⁇ 0 1 become 0, the ZM Prec corrected gait is followed by the same algorithm as the gait this time, and the trapezoidal height of the ZM Prec pattern is set to the acyc2 for the second orbit. When a gait is generated, this gait matches the original normal gait.
- the current gait trapezoid height bcurr at which ⁇ 01 and ⁇ 01 become 0 and the first turning gait trapezoid height bl are obtained, and the obtained bcurr may be finally determined as the trapezoid height of the current gait.
- the dynamic model related to the body posture angle of the robot 1 has linear characteristics like the flywheels FHx and FHy shown in Fig. 12, mu 01 and ⁇ ⁇ are the gait trapezoid height bcurr,
- the first turning gait trapezoid height bl, the difference between the terminal body posture angle of the provisional current time gait and the initial body posture angle of the normal gait err err, the terminal body posture angular velocity of the provisional current time gait and the normal gait It has the following relationship with the initial body posture angular velocity difference v0 err.
- the current gait trapezoid height bcurr the first turning gait trapezoid height bl is determined so that the right sides of Equations 52 and 53 become 0. That is, bcurr and bl are obtained by solving simultaneous equations with the left-hand sides of Equations 52 and 53 set to 0.
- the trapezoidal height of the trapezoidal pattern of the body posture restoring moment ZMP converted value (ZMPrec) of the current time's gait is set to the current time's gait trapezoidal height bcurr obtained above.
- the process proceeds to S720, where the current provisional target ZMP pattern (the provisional target ZMP pattern obtained when the processing exits the repetition loop of S700) is obtained as described above in S710.
- the pattern obtained by adding the body posture restoring moment ZMP conversion value pattern is determined as the target ZMP pattern of the current time's gait. This process is the same as the process of adding a trapezoidal pattern of Aa height to the tentative target ZMP pattern in S710.
- the temporary volume generated in the loop of S7000 is generated by setting the body posture restoration moment ZMP conversion value ZM Prec to 0 (the height parameter of the trapezoidal pattern of ZMP rec to 0.). is there.
- ZM Prec the height parameter of the trapezoidal pattern of ZMP rec to 0.
- the body posture restoring moment ZMP converted value pattern obtained in S 718 generates a body posture angular acceleration for reducing the deviation of the body posture angle from the normal gait to zero.
- the body posture restoring moment obtained in S718 The body posture according to the ZMP converted value pattern
- the dynamic equilibrium condition the combined force of the gravitational force of the mouth pot and the inertial force becomes zero, excluding the vertical component, acting on the target ZMP
- the horizontal position trajectory must be displaced from the provisional current time gait's upper body horizontal position trajectory. Therefore, in the present embodiment, the provisional target ZMP pattern is corrected by Z M Prec so that the body horizontal position trajectory does not need to be shifted from the one finally obtained by the loop of S700.
- the anti-phase arm swing restoration angular acceleration pattern determined here is the period during which the floor reaction force moment vertical component allowable range is sufficiently large in the generation process of the instantaneous gait instantaneous value described later (within the one-leg support period). During this period, the anti-phase arm swing angle trajectory is modified so that it can be connected (closer) to the normal gait.
- This anti-phase arm swing restoration angular acceleration pattern is a trapezoidal pattern, similar to that described in the normal gait generation processing, and is specifically determined as follows.
- the trapezoidal pattern of the anti-phase arm swing restoration angular acceleration of this time's gait is set,
- the time of the vertex (breakpoint) is known (more specifically, the time of the trapezoidal breakpoint is set to the breakpoint time of the target ZMP), and the trapezoid height is set to an unknown value.
- the trapezoidal height (parameter) of the arm swing restoration angular acceleration is required.
- the time when the trapezoidal pattern of the antiphase arm swing restoration angular acceleration starts to rise is Ta, and the time when the trapezoidal pattern returns to 0 is Tb.
- both the anti-phase arm swing angle and the anti-phase arm swing angular velocity are continuously applied to the normal gait. Generally, they cannot be connected. Therefore, in this embodiment, the unknown parameters are determined so that the state of the generated gait gradually approaches the state of the normal gait over a plurality of steps.
- the anti-phase arm swing restoration angular acceleration pattern in one gait is complicated, the number of unknown parameters is set to two or more, and the anti-phase arm swing angle and anti-phase arm swing at the end of the gait this time. Both the angular velocity and the gait may be connected continuously to the normal gait, but the antiphase arm swing restoration angular acceleration pattern may fluctuate too much zigzag.
- the difference between the terminal antiphase arm swing angle of the provisional current time gait and the initial antiphase arm swing angle of the normal gait obtained by setting the trapezoid height of the antiphase arm swing restoration angular acceleration pattern to 0 in S702. , And call it Sazerr. Further, a difference between the terminal antiphase arm swing angular velocity of the provisional current time gait and the initial antiphase arm swing angular velocity of the normal gait is obtained, and this is defined as v 0 zerr.
- the current time's gait is generated with the trapezoidal height of the antiphase arm swing restoration angular acceleration pattern as a certain value bzcurr, and that the first turning gait is subsequently generated by the same algorithm as the current time's gait. I do.
- the anti-phase arm swing restoration angular acceleration pattern of the first turning gait is the anti-phase arm swing restoration angular acceleration pattern (the height of the azcycling gait obtained in S314 in FIG. 24).
- the gait generated in this way is called an anti-phase arm swing restoration angular acceleration corrected gait, and its end (end of the first turning gait) has the anti-phase arm swing angle and angular velocity of 0 zl and v, respectively.
- the original normal gait obtained at the completion of the subroutine processing for obtaining the initial state of the normal gait in S024 (normal gait initial antiphase arm swing angle finally determined in S314)
- the angular velocity as the initial values, and the antiphase arm swing restoration angular acceleration pattern as the pattern (the trapezoidal pattern in Fig. 36 with the height of azcycl) obtained in S314 (the normal gait).
- the turning phase gait end antiphase arm swing angle and angular velocity are ⁇ ⁇ ⁇ ⁇ zlorg and V ⁇ zlorg, respectively.
- ⁇ ⁇ zl ⁇ ⁇ zl— ⁇ zlorg... Equation 1 0 5 1
- ⁇ zl and ⁇ ⁇ are the difference between the anti-phase arm swing angle between the anti-phase arm swing restoration angular acceleration corrected gait and the original normal gait when the gait is generated up to the end of the first turning gait. It means the difference in angular velocity. If ⁇ > zl and ⁇ ⁇ become 0, following the anti-phase arm swing restoration angular acceleration corrected gait, the trapezoidal height of the anti-phase arm swing restoration angular acceleration pattern is calculated using the same algorithm as this time's gait. When the second turning gait is generated as azcyc2, this gait matches the original normal gait.
- the current gait trapezoid height bzcurr and the first turning gait trapezoid height bzl at which ⁇ 0 zl and ⁇ V 0 zl become 0 are determined, and the obtained bzcurr is finally determined as the trapezoid height of the current gait. Just do it.
- lzl and ⁇ are the gait trapezoid height bzcurr, 1Turning gait trapezoid height bzl, terminal antiphase arm swing angle of temporary gait and initial antiphase arm swing of normal gait 2
- ⁇ ⁇ zl czll * bzcurr + czl2 * bzl + 9zerr + ezl * v0zerr
- Equation 1 0 5 3
- czll, czl2, cz21, cz22, ezl, ez2 are the parameters of the current time's gait, the gait cycle of the first turning gait, and the parameters of the antiphase arm swing restoration angular acceleration pattern This is a coefficient that is uniquely determined by
- the coefficients czll, czl2, cz21, cz22, ezl, and ez2 of the equations 1052 and 1053 are the gait of this time, the gait cycle of the first turning gait, and the antiphase arm swing recovery angle. It can be obtained based on acceleration pattern parameters (especially parameters related to time).
- Bzcurr and bzl can be obtained by solving the simultaneous equations with the left-hand side of 105 and Equation 105 set to 0.
- the trapezoidal height of the trapezoidal pattern of the antiphase arm swing restoration angular acceleration of the current time's gait is set to the current time's gait trapezoidal height bzcurr obtained above.
- S030 the subroutine of FIG. 45 is executed. From S 1400 to S 1 41 in FIG. 45, the same processing as S 4 0 to S 4 1 in FIG. 25 is performed. The processing of S1000 to S10018 in FIG. 46, which is a subroutine, is performed.
- the value of the reference body angle at the current time is substituted for the target body angle. Also, the value of the reference arm posture at the current time is substituted for the target arm posture except for the arm posture antiphase arm swing angle and angular velocity.
- the process proceeds to S1002, and it is determined whether or not the current time is during the body posture angle / opposite phase arm swing angle restoration period (the period from time Ta to time Tb). If the determination result of S1002 is N ⁇ , the process proceeds to S104, and if the determination result is YES, the process proceeds to S106.
- the body horizontal acceleration ⁇ , the body angle acceleration i3, and the antiphase arm swing angular acceleration JQ a are calculated during the period during which the body tilt angle and the antiphase arm swing angle are not restored (see FIG. 26).
- the same processing as in S504 to S528) is performed.
- Body angle acceleration (body tilt angle acceleration) i3 is determined.
- the process proceeds to S144, and the body horizontal acceleration and the body posture angular acceleration are integrated to calculate a body horizontal velocity and a body posture angular velocity (body inclination angular velocity). This is further integrated to determine the body horizontal position and body posture (body inclination angle). Note that, in the present embodiment, one body angle of the body posture is determined as a reference body angle.
- the above is the target gait generation processing in the gait generator 100.
- the desired gait is generated as described above.
- the desired body position / posture (trajectory) and the desired arm posture (trajectory) are sent to the mouth-pot geometric model (inverse kinematics calculation unit) 102.
- the desired foot position / posture (trajectory), the desired ZMP trajectory (target total floor reaction force center point trajectory), and the desired total floor reaction force (trajectory) is sent to the composite compliance operation determination unit 104 and also to the target floor reaction force distributor 106.
- the desired floor reaction force distributor 106 distributes the floor reaction force to each of the feet 22 R and 22 L, and determines the desired foot floor reaction force center point and the desired foot floor reaction force. You. The determined target floor floor reaction force center point and the target foot floor reaction force are sent to the composite compliance operation determination unit 104.
- the corrected target foot position / posture (orbit) with mechanism deformation compensation is sent from the composite compliance operation determination unit 104 to the robot geometric model 102.
- the mouth-port geometric model 102 when the target body position / posture (trajectory) and the corrected target foot position / posture (trajectory) with mechanical deformation compensation are input, the legs 2, 2 that satisfy them Calculate the joint displacement commands (values) of the two joints (10 R (L), etc.) and send them to the displacement controller 108.
- the displacement controller 108 controls the displacement of the 12 joints of the lopot 1 by using the joint displacement command (value) calculated by the robot geometric model 102 as a target value.
- the lopot geometric model 102 calculates a displacement specification (value) of the arm joint that satisfies the target arm posture and sends the calculated value to the displacement controller 108.
- the displacement controller 108 uses the joint displacement command (value) calculated by the robot geometric model 102 as a target value to control the displacement of the 12 joints of the arm of the mouth port 1.
- the floor reaction force generated in the robot 1 (specifically, the actual floor reaction force of each foot) is detected by the 6-axis force sensor 50.
- the detected value is sent to the composite compliance operation determination unit 104.
- Erry erry (Specifically, the deviation of the inclination angle of the actual body posture with respect to the vertical direction with respect to the inclination angle of the target body posture with respect to the vertical direction. errx, and the attitude tilt angle deviation in the pitch direction (around the Y axis) is 0 erry) is detected via the attitude sensor 54, and the detected value is sent to the attitude tilt stabilization control calculation unit 112. Sent. In this attitude tilt stabilization control calculation unit 112, the target total floor reaction force center point (target ZMP) for restoring the actual body posture angle of the robot 1 to the target body posture angle is calculated. The horizontal component of the floor anti-moment is calculated and sent to the composite compliance operation determining unit 104.
- target ZMP target total floor reaction force center point
- the compensation total floor reaction force moment horizontal component Mdmdxy is determined by the following equation using, for example, a PD control law.
- the body posture inclination angular velocity deviation is a time differential value of the body posture inclination angle deviation, and means a deviation of the actual body posture inclination angular velocity with respect to the target body posture inclination angular velocity. More specifically, the body posture inclination angle deviation is obtained from the posture inclination angle deviation in the roll direction (about the X axis) of the body 3 of the mouth pot and the posture inclination angle deviation in the pitch direction (about the Y axis). Vector.
- the angle deviation ⁇ ⁇ of the actual body posture angle deviation generated at the mouth port 1 (specifically, the posture angle deviation in the Y direction (about the ⁇ axis) of the actual body posture angle deviation is 0 errz) is detected via the attitude sensor 54, and the detected value is sent to the stabilization control calculation unit 113.
- the vertical component of the total floor reaction force moment around the rotation compensation is calculated and sent to the composite compliance operation determination unit 104.
- the composite compliance operation determining unit 104 corrects the desired floor reaction force based on the input value. Specifically, the target floor reaction force is modified so that the compensated total floor reaction camo acts around the target total floor reaction force center point (target ZMP).
- the compensation total floor reaction force moment vertical component Mdmdz is determined by the following equation using, for example, a PD control law.
- K 0 bz and K obz are predetermined gains.
- the upper body angular velocity deviation is a time derivative of the upper body angular deviation, and means a deviation of the actual body angular velocity from the target body angular velocity.
- the compensating total floor reaction force moment vertical component Mdmdz is calculated by the formula d 2
- K 0bz K 0bz
- the composite compliance operation determining unit 104 corrects the desired floor reaction force based on the input value. Specifically, the target floor reaction force moment horizontal component is corrected so that the compensation total floor reaction camouflage horizontal component acts around the target total floor reaction force center point (target ZMP). Goals that are dynamically balanced Correct the desired floor reaction force moment vertical component by adding a compensating total floor reaction force moment vertical component to the desired floor reaction force vertical component around the floor reaction force center point (target ZMP).
- the composite compliance operation determination unit 104 is equipped with the above-mentioned mechanism deformation compensation so that the corrected target floor reaction force matches the actual port 1 state and floor reaction force calculated from the sensor detection value etc. Determine the corrected target foot position / posture (trajectory). However, since it is practically impossible to match all the states to the target, a trade-off relationship between them should be given to make them compromise. That is, the control deviation for each target is weighted, and control is performed so that the weighted average of the control deviation (or the square of the control deviation) is minimized. As a result, the actual foot position / posture and the total floor reaction force are controlled so as to substantially follow the target foot position / posture and the target total floor reaction force.
- the gist of the present invention lies in the generation of the gait of the mouth port 1 in the gait generator 100.
- the configuration and operation of the above-described composite compliance operation determination unit 104 and the like are described first by the present applicant. The details are described in Japanese Patent Application Laid-Open Publication No. Hei 11-306661, etc.
- this time's gait parameter is, as described above, the terminal divergence component of the current time's gait and the initial divergence component Q [0] of the normal turning gait in the value viewed from the support leg coordinate system of the current time's gait. Modified to match some Q ".
- the divergent component is the gait generated according to the current time's gait parameters using the dynamics model. It is an index for evaluating whether or not the body horizontal position converges on a steady turning gait.
- the divergent component at the end of the gait this time should be such that the initial divergent component of steady turning ci [0] matches the Q "value seen from the supporting leg coordinate system of the gait this time.
- the divergent component must be defined.
- the divergent component defined by Equation 10 is actually a divergent component that approximately satisfies the above properties.
- the gait when the gait was generated using the dynamics model according to the current time's gait parameters, and when the gait was repeatedly generated continuously according to the normal turning gait parameters, the gait was generated. It can be said that the gait parameters were modified this time so that the upper body horizontal position of the gait converged (approached) to the upper body horizontal position of the normal turning gait.
- the target ZMP pattern in the gait parameters of the current time gait is modified so as to satisfy the condition (the current time's gait approaches the normal gait).
- the trajectory indicated by reference numeral B in FIG. 47 indicates the body horizontal position trajectory generated such that the divergent components coincide at the boundary of the gait as described above.
- the trajectory indicated by the symbol A in the figure is the upper body when the current time's gait is generated so that the horizontal position and speed of the upper body at the boundary with the normal turning gait match, and then the normal gait is repeated.
- the horizontal position trajectory is shown.
- the trajectory indicated by reference sign B generally deviates from the trajectory indicated by reference sign A at the boundary between the current time's gait and the first normal turning gait, but then gradually indicated by reference sign A. It converges on (or approaches) the trajectory and almost coincides with the trajectory indicated by the symbol A in the next normal turning gait period. In this way, even with the gait generation method that matches only the divergent component at the gait boundary, the gait can be prevented from diverging as in the gait generation method that matches both the position and speed at the gait boundary.
- the example indicated by reference numeral C in the figure shows an example in which the trajectory is generated without considering them. In such cases, the generated trajectories will diverge over time.
- the target ZMP pattern may be complicated, and several parameters may be adjusted so that both the position and speed match. There is a risk of zigzag.
- the divergence component also coincides if both the position and velocity are matched, so the method of matching both position and velocity can be said to be a special case of the method of matching the divergence component.
- the gait when the gait is generated according to the current gait parameters using the dynamic model, and the gait is repeatedly generated according to the normal turning gait parameters continuously as it is, It can be said that the gait parameters were modified this time so that the generated body posture angle of the gait converged (approached) or matched the body posture angle of the normal turning gait.
- the first turning gait immediately after this time's gait needs to be a gait corrected by the first turning gait trapezoid heights bl and bzl obtained as described above.
- the gait combining the current time gait and the first turning gait is regarded as the current time gait
- the gait parameters are adjusted so that the upper body posture angle of the gait converges (approaches) or coincides with the constant body gait angle of the normal gait consisting of the second turning gait and the first turning gait. It can be said that the night was corrected.
- the floor reaction force horizontal component allowable range can be set independently for the front-rear direction (X-axis direction) component and the left-right direction (Y-axis direction) component for easy understanding.
- X-axis direction front-rear direction
- Y-axis direction left-right direction
- a gait that is harder to slip is generated when expressed by a relational expression in the left-right direction.
- a so-called friction circle may be set as an allowable range as in the following equation.
- Equation 59 Fz is the desired floor reaction force vertical component is the friction coefficient, and ka is a positive constant of 1 or less.
- the allowable range of the pair of the floor reaction force horizontal component and the floor reaction camouflage vertical component is changed. You may make it set. As the horizontal component of the floor reaction force increases, the friction limit of the vertical component of the floor reaction force moment decreases. Also, as the vertical component of the floor reaction force moment increases, the friction limit of the floor reaction force horizontal component decreases. Therefore, in consideration of this, it is better to set the allowable range of the combination of the horizontal component of the floor reaction force and the vertical component of the floor reaction camouflage to set an allowable range closer to the actual friction limit characteristics. Can be. Specifically, an allowable range may be set for the weighted average of the absolute value of the floor reaction force horizontal component and the absolute value of the floor reaction force moment vertical component.
- two motion modes in order to set the floor reaction force horizontal component and the floor reaction force moment horizontal component around the target ZMP to appropriate values, two motion modes, a body tilt mode and a body translation mode, are used.
- the exercise mode is used, other exercise modes may be used.
- one of the motion modes is a motion mode that does not generate the floor reaction force horizontal component.
- any floor can be used. This is because the reaction force horizontal component and the floor reaction force moment around the target ZMP can be generated.
- an exercise mode other than the exercise mode for changing the body posture may be used.
- a motion mode in which the tip positions of the left and right arms are swung in the same direction, or a motion mode in which the position of the foot that is not touching the ground (present in the air) is perturbed may be used.
- the perturbation should be returned to substantially zero just before the nourishment land so that the landing position does not change. It is needless to say that the exercise mode in which the tip positions of the right and left arms are swung in the same direction and the inverted arm swing mode may be used in combination.
- At least two of the selected motion modes need to have different generation ratios between the floor reaction force horizontal component and the floor reaction force moment around the target ZMP due to the motion mode. Otherwise, there is generally no solution to the system of equations (the behavior of each motion mode cannot be uniquely determined).
- the upper body rotation mode instead of the antiphase arm swing mode.
- the upper body rotation mode is used, the upper body 3 of the robot 1 is moved to the waist (for example, the lower part of the member 54 in FIG. 1) and the upper part thereof (for example, FIG. 1). 5)
- the upper part should be provided so that the upper part can rotate in one direction (for example, around the trunk axis of the upper body 3) with respect to the part near the waist. Is preferred. By doing so, the upper part of the upper body 3 can be rotated without affecting the posture of the legs 2, 2, and the vertical component of the floor reaction force moment can be adjusted.
- arms, upper body A motion mode for displacing other parts may be used.
- a mode in which the ends of both legs are moved back and forth may be used.
- the anti-phase arm swing mode and the upper body single rotation mode may be used together.
- the upper body single rotation mode and the antiphase arm swing mode are modes in which the floor reaction force moment vertical component is generated so that the total body weight position does not change (in other words, without generating the floor reaction force horizontal component). There is, however, a movement in which the position of the center of gravity changes (in other words, a floor reaction force horizontal component is generated). This is because the floor reaction water can be adjusted by combining with the upper body translation mode.
- the following model may be used in addition to the dynamic model used in the above reference example (the dynamic model in FIG. 12).
- a nonlinear model in which mass points are set for multiple links.
- Each link in this model may have an inertia (inertia moment).
- the mass point shown in FIG. 12 is a partial model representing the relationship between the resultant force and the body translational movement.
- 1 4 4 is a partial model showing the relationship between the resultant force and the body posture rotational movement.
- the model used in each process may be the same, or the model may be appropriately used depending on the process.
- the requirement for the dynamic accuracy of the normal time's gait is smaller than that of the current time's gait. Therefore, for example, in the gait generation processing this time, while using the dynamic model of FIG. 12 (3 mass points + flywheel model), the generation processing of the normal gait (especially S408, FIG.
- the block diagram, the flowchart, the algorithm, and the like may be equivalently modified such as changing the order of the arithmetic processing, and a one-pass filter may be appropriately introduced.
- the initial state of the normal gait (mainly the initial upper body horizontal position velocity, initial upper body vertical position velocity, Angle / angular velocity) instead of using the above method to calculate various normal gait parameters in advance.
- the relationship between the normal gait parameters and the initial state of the normal gait is mapped or approximated and stored, and during actual movement, it is mapped or approximated.
- the initial value of the normal gait may be determined based on the above relationship.
- the function obtained by combining the above-described relationship mapped or approximated and the function f may be mapped or approximated and stored.
- a function for directly calculating the divergent component of the normal gait from the normal gait parameters consisting of the above-mentioned foot trajectory parameter, floor reaction force vertical trajectory parameter, etc. You may.
- a normal gait is generated in advance for each set of a plurality of typical normal gait parameters, and the initial state of the normal gait for each set of normal gait parameters (Fig. 13), and a map showing the relationship between the normal gait parameters of each group and the initial state of the normal gait is created in advance.
- the initial state of the normal gait may be obtained from the determined set of normal gait parameters by selection or interpolation based on the map.
- the target ZMP parameter of the current time's gait is modified as a method of modifying the current time's gait to connect to (close to) the normal gait, but other parameters may be modified.
- the trajectory of the free leg of the gait this time may be changed in the air. If, for example, the horizontal position of the torso is likely to shift rearward in the normal gait, the free leg is quickly moved forward after leaving the floor to shift the center of gravity of the free leg forward. By doing so, the horizontal position of the upper body in order to satisfy the target ZMP must be accelerated further forward. As a result, the upper body horizontal position At the end of the current time's gait, it moves further forward and can match the normal gait.
- the cycle of the gait may be corrected this time. For example, if the horizontal position of the torso is likely to deviate behind the normal gait, the gait cycle may be increased this time. This is because by increasing the cycle of the gait this time, the time required for the horizontal position of the upper body to move increases, and the gait can move forward accordingly.
- the gait end body torso horizontal position is almost proportional to the amount of correction of the target ZMP. Changes, the number of times of searching for the appropriate value can be reduced.
- the horizontal position of the body at the end of the gait changes considerably non-linearly with respect to the correction. It requires a large number of searches.
- the target ZMP parameter of the current time's gait was corrected as a method of correcting the current time's gait to connect to (close to) a normal gait.
- the correction amount (the correction amount a in FIG. 34) of the target ZMP parameter may be excessive. For example, if a hopping gait on the spot gives a request to move quickly at high speed (running request), the target ZMP parameter is set to connect to (close to) a high-speed normal gait (normal running gait). In the evening, it will be necessary to shift the direction of travel extremely backward. In this case, it is desirable to correct the gait parameters other than the target ZMP parameter as described above. However, in this case, in fact, it was not possible to request the rapid acceleration itself, so the requested value itself may be modified.
- the normal gait that satisfies the request (required parameter) is determined according to the procedure described above, and when the current gait parameter is determined to lead to this normal gait, the target ZMP trajectory of the current gait is stabilized. Determine if the margin is too low. If the stability margin becomes too small (when the target ZMP deviates from the so-called support polygon or the target ZMP is located near the end of the support polygon), the request should be modified. Just do it.
- the allowable range of the gait acceleration / deceleration (the next gait initial speed—the current gait initial speed) or the current gait cycle) in advance and receive the request (request parameter related to the gait period).
- the degree of acceleration / deceleration corresponding to the request may be obtained, and if the obtained acceleration / deceleration exceeds the allowable range, the request may be modified so that the required acceleration / deceleration falls within the allowable range.
- the above ⁇ , ⁇ Fp, ⁇ r, AFr, ⁇ Maz and ⁇ Mbz may be obtained analytically by dynamics calculation when a simple dynamics model is used as described above.
- the floor reaction force when the upper body 3 is accelerated by a small amount in the body translation mode or when the body 3 is accelerated by a small amount in the body tilt mode is determined, and acceleration is not performed.
- the difference from the floor reaction force in this case can be obtained, and the difference can be obtained by dividing the difference by the minute amount.
- the average value of each of ⁇ , AFp ⁇ , ⁇ Fr, AMaz, ⁇ Mbz, and ⁇ Mp ⁇ Mr in a standard gait may be determined in advance and used. Since ⁇ , ⁇ Fp ⁇ r ⁇ Fr, ⁇ Maz, AMbz, and ⁇ vary depending on the state (posture and rate of change), the accuracy is slightly lower than the method obtained for each state at each moment. However, when a model more complicated than the above model is used, the amount of computation can be significantly reduced.
- the trapezoid height bzcurr of the antiphase arm swing restoration angular acceleration pattern of the gait The following method may be used for the determination. .
- the antiphase arm swing angle and angular velocity of the antiphase arm swing restoration angular acceleration corrected gait (see S722 in FIG. 42) at the end of the current time gait are set to 0 zcurr v ⁇ zcurr, respectively.
- a 0 zcerr and ⁇ v 0 zcerr are the difference between the anti-phase arm swing angle and the angular velocity of the normal gait.
- the gait cycle is defined as the interval, the difference between the terminal antiphase arm swing angle of the provisional current time gait, the initial antiphase arm swing angle of the angular velocity and the normal gait, and the angular velocity 0 zerr, ⁇ ⁇ zerr are the previous state, bzcurr ⁇ Zcerr, ⁇ ⁇ zcerr is set as the current state, and a state equation of a discrete system is established, and A 0 zcerr, ⁇ v 0 zcerr is converged to 0 using modern control theory, etc.
- a feedback rule may be determined, and bzcurr may be calculated based on the feedback rule.
- the anti-phase arm swing restoration angular acceleration of the current time's gait and / or the normal gait) 3 arec is not a trapezoidal pattern, but at each moment, the target anti-phase arm swing angle ⁇ angular velocity and the reference anti-phase arm swing angle ⁇ angular velocity
- the value of the reversal phase armrestoring angular acceleration; 8 arec at each instant may be determined based on the difference between the values, such that the difference converges to 0, using a state feedback rule or the like.
- the anti-phase arm swing restoration angular acceleration ⁇ arec of the current time's gait is not trapezoidal pattern, and at each moment, based on the target antiphase arm swing angle and angular velocity of the current time's gait, these are the initial
- the antiphase arm swing recovery angle acceleration ⁇ arec at each instant may be determined using the state feedback rule or the like so as to approach the antiphase arm swing angle and the angular velocity.
- the translational floor reaction force is used.
- the floor parallel component parallel to the floor
- the allowable range of the frictional force or the allowable range of the overall center of gravity acceleration of the floor parallel component (this is proportional to the frictional force).
- the floor reaction force moment vertical component can be calculated by the equation Since it can be converted to frictional force normal direction moment, instead of floor reaction force moment vertical component allowable range, floor reaction force moment normal direction component, that is, floor friction force normal direction moment allowable range
- the allowable range may be set.
- Floor friction force normal moment floor reaction force moment vertical component * cos (0 f)
- Equation 1 07 2 The parameters of the current time's gait are determined not only when the previous time's gait is completed as described in the above reference example, but also described in PCT Publication No. WO 02, 40224 by the present applicant. Like this, even if the gait is being generated, 4 009472
- gait modification redetermining the current time's gait parameters
- the gait that is not to be corrected or the gait that has been temporarily corrected the gait in search
- Gait that does not completely satisfy the search completion condition the gait boundary condition deviation must be within the allowable value
- An appropriate (not provisional) corrected gait may be provided by the end of your cycle.
- the corrected target ZMP trajectory and the desired floor reaction force vertical component trajectory are continuous, and they do not change rapidly after a short time. There is almost no problem, just a slight jagged orbit.
- the target floor reaction force vertical component may be set in a pattern as shown in FIG. 50, for example, instead of the one shown in FIG.
- the floor reaction force vertical component trajectory is set to a trapezoidal shape that is convex on the increasing side of the floor reaction force vertical component during the two-leg support period, and is convex on the decreasing side of the floor reaction force vertical component during the one-leg support period. Set to trapezoidal.
- FIGS. 53 to 58 a first embodiment of the present invention will be described with reference to FIGS. 53 to 58 based on the above-described reference examples and modifications thereof.
- this embodiment In the description of the embodiment, the same components or the same functional portions as those in the reference example are denoted by the same reference numerals as those in the reference example, and description thereof is omitted. Particularly, in the present embodiment, as described above, the matters described with reference to FIGS. 1 to 3 and FIGS. 5 to 12 are the same as those in the reference example.
- the actual body posture angle deviation (the vertical angle of the body 3) is the difference between the target body posture angle and the actual body posture angle.
- the target gait is also modified to make the inclination angle deviation with respect to the direction and the deviation of the single angle) and / or the rate of change thereof close to zero.
- it dynamically balances with the target gait according to the angle component and Z or its angular velocity of the actual body posture deviation (the resultant force of the inertia and the gravity of the motion of the target gait is around the target ZMP).
- the vertical component of the floor reaction force moment around the ZMP is also corrected.
- FIG. 53 is a block diagram illustrating a functional configuration of the control unit 60 according to the present embodiment.
- the points of the functional configuration of the control unit 60 according to the present embodiment that are different from those of the reference example (FIG. 4) will be described.
- the compensating total floor reaction force moment horizontal component Mdmdxy calculated in the posture tilt stabilization control calculation unit 112 is input to the compensating total floor reaction force moment horizontal component distributor 110.
- the compensating total floor reaction cam component horizontal component distributor 110 converts the compensation total floor reaction force moment horizontal component Mdmdxy into the target floor reaction force moment horizontal component for compliance control and the model operation floor reaction force moment horizontal component. And distributed to.
- the posture inclination stabilization control calculation unit 1 1 2 and the compensation total floor reaction force moment horizontal component distributor 1 110 allow the target floor reaction force moment horizontal for compliance control to be obtained.
- the component and the model-operated floor anti-camoment horizontal component are determined. 4 009472
- the model operation floor reaction force moment horizontal component is determined by the following equation.
- the floor reaction force moment horizontal component allowable range is determined by the gait generator 100 as described later.
- the target floor reaction force moment horizontal component for compliance control is determined by the following equation.
- the floor reaction force moment horizontal component is determined such that the difference between the target floor reaction force moment horizontal component for compliance control and the model operated floor reaction force moment horizontal component is equal to Mdmdxy.
- FIG. 54 is a block diagram showing the compensation total floor anti-camouflage horizontal component distributor 110 performing the above-described operation.
- the compensation total floor reaction force moment vertical component Mdmdz (see the above equation d26) determined in the stabilization control calculation unit 113 in the same manner as in the above-described reference example. Is input to the model operation floor reaction force moment vertical component determiner.
- the model operation floor reaction moment vertical component determiner 11 determines the model operation floor reaction cam lead S component based on the compensated total floor reaction moment vertical component Mdmdz. In other words, based on the one-body deviation of the actual body posture angle deviation, the total compensation is performed by the k-stabilization control calculation unit 113 and the model operation floor reaction force moment vertical component determiner 111.
- the floor reaction force moment vertical component Mdmdz and the model operation floor reaction force moment vertical component are determined.
- model operation floor reaction force moment vertical component determiner 111 the model operation floor reaction force moment vertical component is determined by the following equation.
- the floor reaction force moment vertical component compensation amount allowable range is determined by the gait generator 100 as described later.
- the horizontal floor component and the total floor reaction force moment vertical component Mdmdz of the target floor reaction chamoment for compliance control are sent to the composite compliance operation determination unit 104.
- the model operation floor anti-camoment horizontal component and vertical component are sent to the gait generator 100.
- the sum of the compensation total floor reaction camouflage vertical component Mdmdz and the model operation floor reaction force moment vertical component is used as the target value for compliance control. It may be sent to the sensing operation determining unit 104.
- the composite compliance motion determining unit 104 is configured to make the gait generator 100 generate the target gait generated by the gait generator 100 while following the motion of the lopot 1 with the motion of the target gait generated by the gait generator 100.
- the target floor reaction force moment for compliance control is added to the floor reaction force so that the actual floor reaction force approaches the target total floor reaction force corrected by adding the horizontal component of compensation control moment and the compensation total floor reaction moment vertical component Mdmdz.
- the corrected target foot position / posture (trajectory) with mechanical deformation compensation is determined by correcting the flat position / posture.
- the gait generator 100 calculates the floor reaction force moment horizontal component around the target ZMP determined in the gait generator 100 by the model operation floor reaction force model.
- the target gait movement (especially the body position / posture trajectory) is generated using a dynamic model so that the horizontal component is obtained. Further, the gait generator 100 is generated assuming that the model operation floor reaction force moment is 0.
- Model reaction floor reaction cam component vertical component is added to the target floor reaction force moment vertical component around the target total floor reaction force center point (target ZMP) that is dynamically balanced with the desired target gait (temporary target gait). Correct the gait of the desired gait (especially the fl ⁇ g swing trajectory) so that the gait occurs.
- control unit 60 is the same as that of the reference example.
- desired gait generated in the reference example is obtained when the model operation floor reaction force moment horizontal component and the model operation floor reaction force moment vertical component are constantly set to 0. It is the same as the generated desired gait.
- the operation (gait generation processing) of the gait generator 100 in the first embodiment will be described below in detail with reference to FIG. From S310 to S308, the same processing as that of S010 to S028 in FIG. 13 of the reference example is performed.
- the value obtained by dividing the floor reaction force moment horizontal component by the floor reaction force vertical component represents the deviation of the ZMP (floor reaction force center point) from the target ZMP. Therefore, by dividing the floor reaction force moment horizontal component allowable range by the floor reaction force vertical component, the parameters of the ZMP allowable range (floor reaction force center point allowable range) converted to the floor reaction force center point should be set. It may be.
- the floor reaction force moment vertical component is assumed to be the floor reaction force moment vertical component allowable range for gait generation.
- the compensation amount of the floor reaction chamoment vertical component that can be added to the floor reaction force moment that generates the movement of the desired gait Means. Therefore, unless the floor reaction force moment vertical component allowable range for gait generation is set sufficiently narrower than the actual friction limit, the floor reaction force moment vertical component compensation amount allowable range cannot be set wide.
- the allowable range of the floor reaction force moment vertical component compensation amount may be set to be similar to the allowable range of the floor reaction camouflage vertical component for gait generation (see FIG. 41 described above).
- the floor reaction force moment vertical component compensation amount allowable range for compliance control is set in a region where the upper limit value is 0 and the lower limit value is 0.
- the parameters that define the allowable range of the floor reaction force moment horizontal component around the ZMP and the floor reaction force moment vertical component compensation amount in S3003 are set as the target for compliance control.
- the process proceeds to S3302, where the instantaneous gait instantaneous value is determined this time.
- the instantaneous gait instantaneous value is determined so that the model operation floor anti-chamoment horizontal component determined according to the above-described equation d27a is generated around the target ZMP.
- the instantaneous gait value is determined according to the flowchart of FIGS. 57 and 58. That is, in S3003, first, the processing from S340 to S3141 in FIG. 57 is executed. The processing from S340 to S3141 is exactly the same as the processing from S140 to S141 in FIG. 45 described above.
- the process proceeds to S3412, and the parameters of the floor reaction force moment horizontal component allowable range and the floor reaction force moment vertical component compensation amount range for compliance control determined in S3030 in FIG. Based on the evening, the instantaneous values of the floor reaction camouflage horizontal component allowable range [Mxymin, Mxymax] and the floor reaction force moment vertical component compensation amount allowable range [Mzcmin, Mzcmax] at the current time (current time t Value) is required.
- the calculated floor reaction force moment horizontal component allowable range is sent to the compensating total floor reaction force moment horizontal component distributor 110 (see FIG. 53). Then, the current value (the value at the current time t) of the model operation floor reaction force moment calculated by the above-mentioned formula d27a by the distributor 110 is provided to the gait generator 100. '
- the obtained floor reaction force moment vertical component compensation volume range is sent to the model operation floor reaction force moment vertical component determiner 111 (see FIG. 53).
- the current value (the value at the current time t) of the model operation floor reaction force moment vertical component calculated by the model operation floor reaction force moment vertical component determiner 1 11 1 according to the above equation d26b is used as the gait generator. Given to 100. Next, the processing of the gait generator 100 proceeds to S 3 4 1 4, 2
- the body horizontal acceleration and the body posture inclination angle acceleration of this time's gait were set so that the model operation floor reaction force moment horizontal component given from the anti-chamoment distributor 1 110 was generated around the target ZMP. Is determined. However, so that the floor reaction force horizontal component Fx does not exceed the floor reaction force horizontal component allowable range [Fxmin, Fxmax] determined in S3410, the body horizontal acceleration and the body posture angular acceleration (body tilt) Angular acceleration) is determined.
- the horizontal component of the moment acting around the target ZMP is the resultant of the inertial force of the motion of the mouth port 1 and the gravity of the motion of the mouth port 1 becomes the moment in which the sign of the model operation floor reaction force moment horizontal component is inverted.
- the body horizontal acceleration and body posture angle acceleration (body inclination angle acceleration) of the gait are determined this time.
- the force with the sign of the inertial force horizontal component reversed is the floor reaction force horizontal component allowable range.
- the body horizontal acceleration and the body posture inclination angular acceleration are determined so as not to exceed [Fxmin, Fxmax].
- S3414 specifically, the body horizontal acceleration and the body posture angular acceleration are determined according to the flowchart shown in FIG. In this flowchart, the same processes as those in FIG. 26 are performed except for S3104 and S3130. Unlike S504 and S530 in Fig. 26, S3104 and S3130 move from the previous instantaneous gait state of mouth port 1 (gait state at time k-1) to the upper body inclination.
- the angular acceleration of the mode is set to 0 (more precisely, the angular acceleration of the body tilt mode is made to match the reference body posture angular acceleration) and the motion of the body translation mode is set to robot 1,
- the body horizontal acceleration required to generate the model operation floor reaction force moment horizontal component around the target ZMP at (time k) (CK tmp for S3104, and (3 ⁇ 4) for S3130) .
- correction amount of the antiphase arm swing angular acceleration corresponding to the model operation floor reaction force moment vertical component) 3 aadd is obtained by the following equation.
- the target in the gait generator 100 Gait generation processing is performed, and the instantaneous values of the desired body position / posture, the desired foot position / posture, the desired arm posture (including the antiphase arm swing angle), the desired ZMP, and the desired total floor reaction force are sequentially determined. Is output.
- the target total floor reaction force may output only the components necessary for compliance control.
- the target ZMP is included in the target total floor reaction force, but it is particularly important, so it was daringly listed as an output.
- the model operation floor reaction force moment horizontal component is not output as the target floor reaction force.
- the target floor reaction force that aims to make the horizontal component of the floor reaction camoment around the target ZMP zero (the target floor reaction force that satisfies the target ZMP in the original sense) ) Is output from the gait generator 100.
- the floor reaction force moment vertical component of the current time's gait corrected in S3034 is output from the gait generator 100 as a target value.
- the motion of the target gait is generated so that the model operation floor anti-chamoment horizontal component is generated around the target ZMP, and the floor reaction force of the actual Is controlled so that the model operation floor reaction force moment horizontal component is not added. Therefore, the unbalance between the desired gait movement and the floor reaction force is generated by the horizontal component of the subtraction model operation floor reaction force moment.
- the sign of the model operation floor reaction force moment horizontal component can be expressed as the effect of converging the difference between the body posture inclination angle of the actual robot 1 and the desired gait body posture inclination angle to 0. This is equivalent to applying the inverted floor reaction force moment horizontal component to actual lopot 1.
- the model operation floor reaction force moment horizontal component is determined appropriately.
- the real robot 1 can be converged to the corrected target gait (a gait that converges the difference between the body posture inclination angle of the actual robot 1 and the desired gait body posture inclination angle to 0). it can. That is, the posture inclination of the actual mouth port 1 can be stabilized.
- the sum of the moment in which the sign of the model operation floor reaction force moment horizontal component is inverted and the target floor reaction force moment horizontal component for compliance control is the total tilt restoring force (actually, robot 1).
- the effect 4 is that the body reaction acceleration of the body translation mode and the body posture inclination angular acceleration are set so that the floor reaction force horizontal component does not exceed the floor reaction force horizontal component allowable range.
- the locomotive 1 moves on a floor where a large floor reaction force horizontal component cannot be generated, such as immediately before leaving the support leg side leg 2 or immediately after landing in a running gait, or on a floor with a small coefficient of friction In this case, the slip of the mouth port 1 can be prevented.
- the allowable range of the floor reaction force horizontal component is set to 0 at a time when the translational force vertical component of the floor reaction force is 0, that is, at a time when both legs are not in contact with the ground, and thus,
- the algorithm of the first embodiment automatically restores the posture inclination depending on the body tilt mode without depending on the body translation mode, and does not depend on the frictional force between the floor and the sole. Posture recovery is performed at the same time. Therefore, even during this period (the mid-air period), unlike the method of simply correcting the upper body translation mode, the posture inclination restoration effect works effectively. At this time, since the floor reaction force horizontal component is generated to be 0, the overall center of gravity horizontal acceleration of the gait is 0.
- the model-operated floor reaction force moment horizontal component is not output as the target floor reaction force for compliance control.
- the floor reaction force moment horizontal component around the target ZMP becomes zero for compliance control.
- a desired floor reaction force that aims to be obtained is given from the gait generator 100. Therefore, the floor reaction force control by the compliance control can be appropriately performed without hindering the floor reaction force control by the compliance control. More specifically, it is possible to prevent or suppress the occurrence of a problem that the original contact property of the foot 22 is deteriorated and the bottom surface of the foot 22 floats.
- the target floor reaction force moment horizontal component for compliance control around the target ZMP is determined so as not to exceed the floor reaction force moment horizontal component allowable range.
- the motion of the desired gait is generated so that the model operation floor reaction force moment vertical component is additionally generated around the target ZMP, while the actual floor reaction force of the robot 1 is calculated by the gait generator.
- the combined compliance is defined as the target value that is obtained by adding the compensated total floor reaction force moment vertical component Mdmdz to the desired floor reaction force moment vertical component that balances the desired gait with the model operation floor reaction force moment vertical component added by 100
- the actual floor reaction force is controlled by the control so as to approach the target value.
- Mdmdz becomes larger, the model operation floor anti-camo-motion vertical component in the opposite direction to Mdmdz is added to the desired gait.
- the actual mouth port 1 can be adjusted so that the spin does not occur and the corrected target gait (the upper body posture of the real robot 1 and z Alternatively, it is possible to converge to a gait in which the difference between the body posture angle angular velocity and the body posture angle angle Z of the target gait or the body posture angle angular velocity converges to zero. In other words, one rotation of the actual robot 1 can be stabilized.
- the vertical moment component Mdmdz of the total floor reaction force moment becomes the total restoring force.
- the compensation total floor reaction force moment vertical component Mdmdz is determined based on the feed-pack control law so that the angular deviation and Z or angular velocity deviation approach 0, so that the control stability of the angular deviation is guaranteed. Meanwhile, the angular deviation and the angular velocity deviation can be reduced to zero.
- the model operation floor reaction force moment vertical component can take any value ignoring the allowable range (or friction limit) of the floor reaction force moment vertical component.
- a single rotation restoring force can be generated.
- the target floor reaction cam vertical component is the sum of the floor reaction force moment vertical component allowable range and the floor reaction force moment vertical component compensation amount allowable range. It is determined not to exceed the range, so just before leaving the support leg side leg 2 in the running gait or landing 4 009472
- the above-described algorithm of the present embodiment automatically restores a single rotation that depends on the anti-phase arm swing mode without depending on the actual floor reaction force moment vertical component.
- One rotation restoration is performed without depending on the frictional force between the floor and the sole. Therefore, even during this period (mid-air period), unlike the method of simply correcting the vertical component of the desired floor reaction force moment of compliance control, the one-rotation restoring effect works effectively.
- the gait When the gait is called a modified gait, the gait usually differs from the original gait and the modified gait. Since the original gait is set to approach the normal gait, the corrected gait is usually a gait that does not approach the normal gait.
- the above-mentioned operation 12 is almost the same as the technique of PCT / JP03 / 004435 previously proposed by the present applicant. However, in addition to this, the following operation also occurs in the present embodiment. That is, a normal gait in which the antiphase arm swing angle trajectory of the new current time gait is newly set with the terminal phase of the antiphase arm swing angle and the angular velocity corrected for restoring unidirectional rotation as a new initial state.
- the parameters for the new antiphase arm swing angle trajectory of the new gait are determined so as to asymptotically approach the antiphase arm swing angle trajectory.
- a gait in which stability is guaranteed can be continuously generated.
- the model operation floor reaction camouflage horizontal component becomes 0.
- the model operation floor reaction force moment vertical component compensation amount is the state quantity of the dynamic model shown in Fig. 12 '(for example, the antiphase arm swing angle and angular velocity of robot 1 on the dynamic model, (Angular velocity, angular velocity, etc.).
- the first embodiment described above is an embodiment of the first invention, the second invention, the fourth to ninth inventions, and the fifteenth invention of the present invention.
- the floor reaction force moment vertical component (the floor reaction force moment vertical component to which Mdmdz is added) in the first embodiment is limited, the upper body angle deviation and the noise.
- the body angular velocity deviation is equivalent to the deviation of the state quantity of mouth port 1
- Mdmdz is equivalent to the compensating floor reaction force moment.
- the sum of the floor reaction force moment vertical component allowable range [Mzmin, Mzmax] and the floor reaction force moment vertical component compensation amount allowable range for generating a gait in the first embodiment is the allowable target amount. Equivalent to range.
- the motion component of the instantaneous gait instantaneous value determined in S303 of FIG. 56 corresponds to the temporary instantaneous value of the target motion
- the one corrected in S304 is the instantaneous value of the target motion.
- the dynamic model in the invention corresponds to the dynamic model in FIG.
- the original gait and the corrected gait are generated at the same time.
- the corrected gait is corrected from the original gait to stabilize the body posture (tilt angle and angle) of the actual robot 1. Furthermore, if the corrected gait still has room for generating the floor reaction force moment required for posture recovery by compliance control (if there is room for the floor reaction force moment that can be generated around the target ZMP), Using this margin, it is determined to converge to the original gait as much as possible.
- FIG. 59 is a block diagram illustrating a functional configuration of the control unit 60 according to the present embodiment.
- the compensating total floor reaction force moment horizontal component Mdmdxy obtained by the posture / inclination stabilization control calculation unit 112 is determined by the gait generator 1 0 Input to 0.
- the compensated total floor reaction force moment vertical component Mdmdz obtained by the stabilization control calculation unit 113 is also input to the gait generator 100.
- a compensated total floor reaction moment that determines the model operation floor reaction force moment (horizontal component and vertical component) and the target floor reaction moment for compliance control (horizontal component and vertical component).
- the distributor 120 is incorporated in the gait generator 100, and the target floor reaction force moment for compliance control is output from the gait generator 100 to the composite compliance operation determination unit 104.
- the compensating total floor reaction force moment distributor 120 in the gait generator 100 includes the compensating total floor reaction force moment horizontal component distributor 110 and the compensating floor reaction force moment distributor 110 of the first embodiment.
- model operation floor reaction force moment vertical component determinator 1 1 1 Performs more complicated processing.
- the functional configuration of the control unit 60 is the same as that of the first embodiment.
- FIG. 60 shows a flowchart of the main routine processing of the gait generator 100 in the present embodiment.
- the processing from S 210 to S 210 is the same as the processing from S 110 to S 028 in the main flowchart (FIG. 13) of the reference example. Processing is performed. Note that in the initialization of 'S800' in the flowchart of FIG. 43, which is a subroutine of S028 (in the present embodiment, S2208), the initial state of the current time's gait is the previous corrected gait ( The terminal state of the gait that is finally output by the gait generator 100 is converted to the supporting leg coordinate system this time, and the terminal state of the original gait determined in S20332 described later is , And are not used in S820 of the subroutine of S208.
- the process proceeds to S 2 0 3 2, where the instantaneous value of the original gait (the current fit at time t ) Is determined.
- the original gait is a gait generated so that the horizontal component of the floor reaction force moment about the target ZMP becomes zero.
- the original gait is generated by an algorithm in which a part of the subroutine processing of S3302 in FIG. 56 of the first embodiment is changed. That is, the subroutine processing in S3302 (specifically, the subroutine processing in S3414 in FIG. 57, which is the subroutine processing in S3302), the S3104 and S3104 in FIG. In 130, the horizontal operation of the model operation floor reaction force moment is set to 0 (the horizontal component of the desired floor reaction camouflage around the target ZMP is set to 0), and the body horizontal acceleration atmp is obtained.
- the other processing may be the same as the processing of S3032 in FIG.
- a model horizontal position difference between the models which is a difference between the corrected horizontal gait horizontal position and the original gait horizontal position, is obtained.
- the current value of the corrected gait's upper body horizontal position (the value at time t) has not yet been determined. Therefore, in S2200, the previous value of the corrected gait's upper body horizontal position (the value finally obtained in the control cycle of time t ⁇ 1 ⁇ t) and the previous value of the original gait's upper body horizontal position are calculated.
- the model An interbody horizontal position difference is calculated.
- a model body posture inclination angle difference which is a difference between the corrected gait body posture inclination angle and the original gait body posture inclination angle, is obtained.
- the previous value of the corrected gait body posture inclination angle and the upper body of the original gait are the same as in the case of the calculation process of the body horizontal position difference between the models in S2200.
- the difference between the body posture inclination angles between the models is obtained using the previous value or the current value of the posture inclination angle.
- an anti-phase arm swing angle difference between models which is the difference between the anti-phase arm swing angle of the corrected gait and the anti-phase arm swing angle of the original gait, is determined.
- the previous value of the antiphase arm swing angle of the corrected gait and the antiphase of the original gait are the same as in the calculation of the body horizontal position difference between the models in S2200.
- the anti-phase arm swing angle difference between the models is obtained using the previous value or the current value of the arm swing angle.
- the process proceeds to S 2 206, and based on the body horizontal position difference between the models, the model body horizontal position stabilization floor reaction force module required to converge the difference to 0 Request value Mpfdmd is determined.
- the upper body horizontal position difference between the models diverges.
- the model body horizontal position stabilization floor anti-camouflage required value Mpfdmd is obtained when the body horizontal mode of the corrected gait is returned to the body horizontal position of the original gait in the body translation mode. It has the meaning as a moment obtained by subtracting the floor reaction chamoment that generates the body horizontal acceleration in the upper body translation mode of the original gait from the floor reaction force moment that is generated along with this.
- the required value Mpfdmd of the model body horizontal position stabilization floor reaction force moment is determined, for example, by the following feedback control law.
- the PD control law is used as the feedback control law
- another feedback control law such as PID may be used.
- Mpfdmd Kmp * body horizontal position difference between models
- Equation d 28 Equation d 28 where Kmp and Kmpv are feedback gains (proportional gains, differential gains).
- Kmp and Kmpv are feedback gains (proportional gains, differential gains).
- the process proceeds to S 228, and based on the difference between the body posture inclination angles between the models, the required value of the model body posture inclination angle stabilized floor anti-chamoment required to converge the difference to 0 is Mrfdmd. It is determined.
- the floor reaction force moment that simply generates the body posture tilt angle acceleration of the corrected gait upper body tilt mode is converted to the floor reaction camouflage that generates the body posture tilt angle acceleration of the original gait body tilt mode. If they match, the body posture inclination angle difference between the models will not converge to zero.
- the body posture inclination angle stabilized floor reaction force moment demand value Mrfdmd is calculated when the body posture inclination angle of the corrected gait is returned to the body posture inclination angle of the original gait in the body inclination mode.
- this is a moment obtained by subtracting the floor reaction camo that generates the body posture tilt angle acceleration of the original body gait from the floor reaction force moment generated by this.
- the required value of the model body posture inclination angle stabilized floor anti-chamoment Mrfdmd is determined, for example, by the following feedback control law.
- the PD control law is used as the feedback control law, but another feedback control law such as PID may be used.
- Mrfdmd Kmr * Difference in body posture inclination angle between models
- Equation d 2 9 Equation d 2 9 where Kmr and Kmrv are feedback gains (proportional gains and differential gains).
- the process proceeds to S 2 210, and based on the anti-phase arm swing angle difference between the models, the required value Mafdmd of the model anti-phase arm swing angle stabilization floor anti-camo necessary for converging the difference to 0 is obtained. It is determined.
- An anti-float camouflage that generates anti-phase arm swing angular acceleration in the anti-phase arm swing mode of the original gait simply generates an anti-phase arm swing angular acceleration in anti-phase arm swing mode of the corrected gait , The antiphase arm swing angle between models does not converge to 0.
- the model anti-phase arm swing angle stabilized floor reaction force moment demand value Mafdmd performs the operation of returning the anti-phase arm swing angle of the corrected gait to the anti-phase arm swing angle of the original gait in the inverted phase arm swing mode.
- the anti-phase arm swing angular acceleration in the anti-phase arm swing mode of the original gait is generated. 2004/009472
- the required value Mafdmd of the model antiphase arm swing angle stabilized floor reaction force moment is determined, for example, by the following feedback control law.
- the PD control law is used as the feedback control law, but another feedback control law such as PID may be used.
- Equation d29b where Kar and Kav are feedback gains (proportional gain, derivative gain).
- Kar and Kav are feedback gains (proportional gain, derivative gain).
- an anti-phase arm swing angular acceleration in the anti-phase arm swing mode of the original gait is generated from the vertical component of the floor reaction force moment generated in the anti-phase arm swing mode of the finally determined corrected gait.
- the moment obtained by subtracting the vertical component of the floor reaction force moment is called the model inversion arm swing angle stabilized floor reaction force moment.
- Equation d 30 Considering that this equation d 30 approximately holds and that the vertical component of the floor anti-chamoment changes in proportion to the antiphase arm swing angular acceleration, the model body Determine the horizontal position-stabilized floor reaction force moment to be equal to or as close as possible to the required model body horizontal position-stabilized floor reaction force moment Mpfdmd, and model the model body posture tilt angle stabilized floor reaction force moment.
- the body posture tilt angle stabilization floor reaction force moment It is determined so as to match or be as close as possible to the required value Mrfdmd, and the model antiphase arm swing angle stabilization
- the floor reaction force moment is model antiphase arm swing angle stabilization floor If it is determined to match or be as close as possible to the desired value of the reaction force moment Mafdmd, a model operation floor reaction chamois appropriate for the gait is generated, and the correction is performed while satisfying the following restoration conditions.
- a body horizontal acceleration and body posture inclination angle acceleration of capacity can converge to the extent possible to the body horizontal acceleration and body posture inclination angular acceleration of the original gait, respectively.
- the process proceeds to S 2 212, and the model body horizontal position stabilization floor reaction force moment (body body) is set so as to satisfy the following conditions (referred to as restoration conditions) as much as possible.
- Restoration conditions floor reaction force moment in translational mode and model body posture tilt angle stabilized floor reaction force moment (Floor reaction force moment in body tilt mode) )
- the model antiphase arm swing angle stabilized floor anti-camo Furthermore, to satisfy the definitions of the model body horizontal position stabilized floor reaction force moment, model body posture tilt angle stabilized floor reaction force moment, and model antiphase arm swing angle stabilized floor reaction force moment, The body horizontal acceleration, the body posture inclination angular acceleration, and the antiphase arm swing angular acceleration of the corrected gait are determined.
- restoration conditions shown below the ⁇ there are conditions that number is not compatible with c, that contrary to have high low condition as the priority, causes satisfied with priority number is less condition (satisfied). However, restoration conditions 1, 2, and 3 must be satisfied (established).
- the model body posture inclination angle stabilized floor reaction force moment must match or be as close as possible to the model body posture inclination angle stabilized floor reaction force required value Mrfdmd.
- This condition is a condition for the corrected body posture inclination angle to converge to the body posture inclination angle of the original gait (the originally planned gait).
- the floor reaction force moment must match or be as close as possible to the required model floor horizontal position stabilization floor reaction force moment Mpfdmd.
- the horizontal position of the upper body of the corrected gait is This is the condition for converging to the horizontal position of the upper body of the initially planned gait).
- the processing of S 2 212 to determine the body horizontal acceleration, the body posture inclination angular acceleration, the anti-phase arm swing angular acceleration, etc. that satisfies the above restoration conditions 1 to 6 is specifically described as, for example, It is performed as follows.
- model body horizontal position stabilization floor reaction force moment and the model body posture tilt angle stabilization floor reaction force moment are determined so as to satisfy the above restoration conditions 1, 2, 4, and 5.
- the acceleration and body posture tilt angle acceleration are determined.
- the details of this processing are described in the technique of PCT / JP03Z00435 previously proposed by the applicant of the present invention, and thus the description thereof is omitted here.
- model anti-phase arm swing stabilized floor reaction force moment is determined so as to satisfy the above-described restoration conditions 3 and 6, and the anti-phase arm swing angular acceleration is further determined.
- the process proceeds to S 2 214, and the model operation floor reaction force moment horizontal component is determined by the equation d 30.
- the sum of the model body horizontal position stabilization floor reaction camoment and the model body posture tilt angle stabilization floor reaction camoment obtained in S 2 208 is the model operation floor reaction force moment horizontal component. It is determined.
- the floor reaction force moment around the target ZMP may be directly calculated based on the instantaneous value of the motion of the corrected gait finally determined, and this may be used as the model operation floor reaction force moment.
- the process proceeds to S2216, and the target floor reaction force moment horizontal component for compliance control is determined by the equation d27b. That is, the sum of the compensated total floor reaction force moment horizontal component Mdmdxy and the model operation floor reaction force moment horizontal component obtained in S2124 is determined as the target floor reaction force moment horizontal component for compliance control.
- S2114 in FIG. 61 the processing of S2114 in FIG. 61 is completed, and the process proceeds to S2116.
- the process of S2116 is the same as S3416 of FIG. 57 in the first embodiment, and the current value of the body horizontal position is determined by the second-order integration of the body horizontal acceleration.
- the current value of the body posture inclination angle is determined by the second-order integration of the body posture inclination angle acceleration.
- S2118 The processing of S2118 is the same as S3418 of FIG. 57 in the first embodiment, and the current value of the antiphase arm swing angle is obtained by the second-order integration of the antiphase arm swing angular acceleration. Is determined.
- the state quantity of the dynamic model (or the instantaneous value of the previous or previous gait) is also required.
- Two kinetic models for gait generation are required. In the present embodiment, those dynamic models are the dynamic models shown in FIGS.
- the original gait and the corrected gait are generated in parallel, and the corrected gait is used to stabilize the posture ('inclination angle and single angle) of the actual mouth port 1.
- the corrected gait is used to stabilize the posture ('inclination angle and single angle) of the actual mouth port 1.
- the floor reaction force moment horizontal component and vertical component
- use this margin to make the necessary steps as far as possible. It is trying to converge to the volume. Therefore, in addition to the operational effects of the first embodiment, it is possible to generate a gait close to the originally set gait, that is, close to the originally required gait. Therefore, the preset travel route 09472
- the fact that the corrected body posture inclination angle converges to the body posture inclination angle of the original gait (the originally determined gait) is based on the assumption that the corrected body horizontal position is the original gait (the originally determined gait). Gait) has priority over converging to the horizontal position of the upper body (adjustment of the body translation mode motion as much as possible within a range that satisfies the floor reaction force horizontal component allowable range). Large fluctuations can be suppressed.
- the second embodiment described above is an embodiment of the first to ninth inventions and the fifteenth invention of the present invention.
- the vertical component of the floor reaction force moment (the component of the floor reaction force moment to which Mdmdz is added) in the second embodiment is the limited target amount, the upper body angular deviation and / or the upper body angular velocity deviation. Is the deviation of the state quantity of robot 1, and Mdmdz is equivalent to the compensation floor anti-chamber.
- the floor reaction force moment vertical range allowable range [Mzmin, Mzmax] for the gait generation in the second embodiment corresponds to the allowable range of the amount to be restricted, and the model antiphase arm swing angle stability in the second embodiment.
- the demanded moment Mafdmd corresponds to the model corrected floor reaction force moment in the third invention.
- the motion when the uncorrected floor reaction force moment vertical component is obtained in the subroutine of S2114 in FIG. 61 corresponds to the predetermined provisional instantaneous value in the second invention.
- the dynamic model in the invention corresponds to the dynamic model in FIG. '
- the functional configuration of the control unit 60 is the same as that of the second embodiment, that is, the one shown in FIG. 59.
- the gait generation algorithm executed by the gait generator 100 is different from that of the second embodiment.
- the processing of each unit other than the gait generator 100 is the same as that of the second embodiment.
- FIG. 63 is a block diagram illustrating an outline of a process performed by the gait generator 100 in the present embodiment. With reference to FIG. 63, an outline of the processing of the gait generator 100 will be described below. The outline of the processing described below with reference to FIG. 63 is the same for the fourth to sixth embodiments described later. In the present embodiment and fourth to sixth embodiments described later, the dynamic model in FIG. 12 is referred to as a simplified model.
- the gait generator 100 includes a gait parameter overnight determination unit 100a.
- the gait parameter determination unit 100a determines the value of the parameter of the target gait (parameter for specifying the target gait) or a time-series table. This corresponds to the processing from S355 to S350.
- the parameters determined by the gait parameter overnight determination unit 100a include the desired foot position / posture trajectory, the target arm posture trajectory, the reference body posture trajectory.
- the target ZMP trajectory, the target floor In addition to the parameters that specify the reaction force vertical component trajectory, etc., the parameters that specify the floor reaction force horizontal component allowable range, and the parameters that specify the ZMP allowable range (or floor reaction force moment horizontal component allowable range) And the parameter that defines the allowable range of the vertical component of the floor reaction force moment.
- the floor reaction force horizontal component permissible range and the floor reaction force moment vertical component permissible range set in the present embodiment are the same as those for the simplified model gait set in the processing of S 3 526 described below.
- Step set by S 3 5 3 0 There are two types, one for volume correction.
- the ZMP allowable range (or floor reaction force moment horizontal component allowable range) is only for the full model correction (for gait correction) set in the processing of S350.
- a parameter that defines the allowable range of the ZMP is set in S3530, but this is a parameter that defines the allowable range of the floor reaction force moment horizontal component.
- This is obtained by dividing the floor reaction force moment horizontal component by the target floor reaction force vertical component as described with reference to S300 of FIG. ) Indicates the amount of deviation from the target ZMP.
- the ZMP allowable range set in the present embodiment is set as shown in FIG.
- the details are described in detail in PCT / JP03Z004304, so that further description is omitted here.
- the gait parameters determined by the gait parameter overnight determining unit 100a are input to the target instantaneous value generating unit 100b.
- the target instantaneous value generator 100b is based on the entered gait parameters based on the desired foot position / posture, target ZMP, target floor reaction force vertical component, target arm posture, target overall center of gravity vertical position, target body vertical Position.
- the instantaneous values of the floor reaction force horizontal component allowable range, ZMP allowable range, reference body posture angle and reference antiphase arm swing angle at the current time t are sequentially calculated (generated). Only the target instantaneous value is shown).
- the process of the target instantaneous value generation unit 100b is performed in the process of a flowchart S3532 shown in FIG.
- the target instantaneous values calculated by the target instantaneous value generation unit 100b some of the instantaneous values (specifically, the instantaneous value of the target body vertical position) are provisional values. Yes, and will be corrected later.
- the instantaneous values of the floor reaction force horizontal component allowable range and the floor reaction camoment vertical component allowable range calculated by the target instantaneous value generation unit 100b are calculated using a simplified model. There are instantaneous values for gaits and instantaneous values for gait correction. .
- the target instantaneous value calculated (generated) by the target instantaneous value generation unit 100b (partially a provisional instantaneous value) is input to the full model correction unit 100c.
- the full model correction unit 100c includes a compensating total floor reaction force moment horizontal component Mdmdxy obtained by the attitude tilt stabilization control calculation unit 112 (see FIG. 59) and the Kyo stabilization control.
- the compensated total floor reaction force moment vertical component Mdmdz obtained by the calculation unit 113 (see Fig. 59) is also input.
- the full model correction unit 100c includes a simplified model 100c1 and a full model 100c2 as dynamic models.
- the full model correction unit 100c determines the target body position / posture and the provisional instantaneous value of the antiphase arm swing angle from the input values based on the simplified model 100c1, and further determines The estimated instantaneous values of the body position and the antiphase arm swing angle are corrected using the full model 100c2.
- the full model 100c2 includes either an inverse full model (inverse dynamics full model) or a forward full model (forward dynamics full model), as described later.
- the full model correction unit 100c basically executes the processing of B so as to satisfy the following conditions A1 to A4. That is, the full model correction unit 100 c
- Compensated total floor reaction camouflage horizontal component Mdmdxy and compensated total floor reaction camouflage vertical component Mdmdz are added to the floor reaction force moment that matches the motion of the corrected gait generated by the full model correction unit 100c. Is equal to the floor reaction force moment for compliance control output from the full model correction unit 100c.
- Floor reaction force horizontal component is within the allowable range of floor reaction force horizontal component for gait correction.
- Target floor for compliance control to be generated around target ZMP. In order to satisfy the condition of c within the allowable range of the vertical component,
- the above condition A2 is equivalent to keeping the floor reaction force moment generated around the target ZMP within the floor reaction camoment horizontal component allowable range corresponding to the ZMP allowable range.
- the simplified model 100c1 is a dynamic model that emphasizes computational complexity and ease of behavior analysis rather than dynamic accuracy. However, changes in angular momentum around the center of gravity may be ignored), or inconsistencies (less rigorous) may be used.
- the kinetic model of FIG. 12 (the kinetic model described by the equations 01 to 05) described in the reference example is used as the simplified model 10 ⁇ c 1.
- the full model 100c2 means a different model of the dynamics of the dynamics from the simplified model 100c1. It is desirable that this is a robot dynamic model with higher approximation accuracy than the simplified model 100 c1.
- the dynamic model shown in FIG. 12 is used as the simplified model 100 c 1.
- a dynamic model with high approximation accuracy for example, a robot dynamic model such as a multi-mass model (a model having a mass at each link of mouth port 1) shown in FIG.
- the full model 100 c 2 may set an indignation moment around the mass point.
- the simplified model 1 0 0 c 1 and the full model 1 0 0 c 2 have the same dynamic equation, but differ in the allowable range of the floor reaction force horizontal component and / or the floor reaction camo vertical component, That is, the allowable range for the simplified model gait and the allowable range for the gait correction (for the full model correction) may be different.
- the floor reaction force horizontal component allowable range and floor reaction camoment vertical component allowable range when generating a gait using the simplified model 100 c1 are increased (the friction limit may be exceeded).
- the floor reaction force horizontal component allowable range and floor reaction force moment vertical component allowable range when correcting the gait using the full model 100c2 so that the slip or spin of the port 1 does not easily occur. It may be just a narrow setting.
- the body position / posture and antiphase arm swing angle are input (input).
- the model used to calculate (output) is called the “forward dynamic model” and is based on the target foot position / posture, target body posture, target body position, and antiphase arm swing angle.
- the model used to calculate (output) the floor reaction force (especially, the target ZMP or the floor reaction camoment (horizontal component and vertical component) and the floor reaction force horizontal component around the target ZMP) Called the "inverse dynamics model”.
- the input to the forward dynamics model includes at least the desired floor reaction force, and the input to the inverse dynamics model includes at least Goal exercise is included.
- the full model 100 c included in the full model correction unit 100 c is an inverse dynamics full model (often abbreviated as “inverse full model”) or a forward dynamics full model (often abbreviated as “forward full model”) Is provided.
- inverse full model inverse dynamics full model
- forward full model forward dynamics full model
- the computational complexity of the forward dynamics model tends to be larger than that of the inverse dynamics model.
- the gait generator 100 in the present embodiment executes the processing shown in the flowchart of FIG. 65 to generate a gait.
- the floor reaction force horizontal component allowable range for the full model correction is, for example, the X-axis direction similar to the floor reaction force horizontal component allowable range for the simplified model gait.
- the pattern is set as shown in FIG. 30 based on the floor reaction force vertical component trajectory and Equation 12 above.
- the value of ka * ⁇ in Equation 12 is set as a parameter that defines the floor reaction force horizontal component allowable range for full model correction.
- the allowable range of the floor reaction force horizontal component is surely determined by setting the value of the coefficient ka in Equation 12 smaller than the allowable range of the floor reaction force horizontal component for the simplified model gait. It is desirable to set the range within the limit.
- the allowable range of the vertical component of the floor reaction camoment is set in the same manner as the allowable range of the floor reaction force horizontal component. That is, the pattern is set as shown in FIG.
- the ZMP allowable range is set in the same manner as in the case of the first embodiment described with respect to the setting of the floor reaction force moment horizontal component allowable range in S300 of FIG.
- the ZMP allowable range By multiplying the ZMP allowable range by the target floor reaction force vertical component, it is of course possible to convert the ZMP allowable range into an equivalent floor reaction camouflage horizontal component allowable range.
- S3552 the instantaneous gait instantaneous value (the value at the current time t) is determined using the simplified model (the dynamic model in FIG. 12).
- the processing of S3552 is the same as the processing of S030 of FIG. 13 in the reference example.
- the instantaneous value of the gait at the current time t generated by the processing up to S35332 described above is hereinafter referred to as a simplified model gait instantaneous value.
- the instantaneous value of the simplified model gait is calculated by using the simplified model (the dynamic model shown in FIG. 12) and calculating the combined force between the inertial force and gravity generated by the motion at the mouth port 1. It is determined so that the floor reaction force moment water generated around the ZMP becomes zero (so as to satisfy the dynamic equilibrium condition for the target ZMP).
- the instantaneous value of the body horizontal position and the body posture inclination angle, the instantaneous value of the body vertical position, and the instantaneous value of the antiphase arm swing angle are provisional instantaneous. This value is corrected by the full model correction described later.
- the instantaneous value of the horizontal component of the desired floor reaction force moment around the target ZMP is a force that is constantly 0.
- a target floor reaction force moment horizontal component for compliance control is generated as a target value of the floor reaction force moment horizontal component to be generated.
- the process proceeds to S35536, in which a corrected gait is generated (gait correction) using the full model, and the final instantaneous value of the final desired gait is determined. That is, as described with reference to FIG. 63, the corrected target body position / posture, the corrected target antiphase arm swing angle, and the desired floor reaction force moment (horizontal component and vertical component) around the target ZMP are obtained. Calculation (determination) of the target floor reaction force moment (horizontal component and vertical component) for compliance control is performed.
- the process proceeds to S 358, increases the time t by ⁇ t, returns to S 354 again, and the processes from S 355 to S 358 are repeated.
- the processing of S 356 is a feature of the present embodiment, and the processing will be described in detail below.
- the gait correction method of the device according to the present embodiment is a full model feed forward correction type.
- this method uses the inverse dynamics full model (inverse full model), does not correct the input of the simplified model gait, and uses the perturbation model. 4 009472
- FIG. 66 shows the operation of the gait generator 100 according to the present embodiment, specifically, the gait correction method of S 356 in the mouth chart of FIG. 65.
- FIG. 4 is a functional block diagram to be described.
- the simplified model 200 in FIG. 66 is not only a dynamic model, but also the processing from S3510 to S3352, that is, the instantaneous value of the simplified model gait. Indicates calculation (decision) processing. Therefore, in FIG. 66, the part after the simplified model 200 corresponds to the processing of S355.
- the processing for determining the instantaneous values of the floor reaction force horizontal component allowable range and the ZMP allowable range for gait correction (for full model correction) is performed by using the reference numeral S3553 in the flowchart of FIG. Is shown.
- the block diagram Since the actual processing is executed by one computer, the block diagram is discretized and then executed sequentially from the upstream side to the downstream side (gait output side) of the block diagram. However, as the feedback amount returning to the upstream side, the value (state quantity) calculated in the previous control cycle (time t- ⁇ t with respect to current time t) is used. Hereinafter, the value calculated in the previous control cycle (time t ⁇ t) is abbreviated as the previous value.
- the target body posture angle of the simplified model gait obtained as described above (hereinafter referred to as the simplified model body posture angle.
- Target body horizontal position (hereinafter referred to as simplified model horizontal position), target center of gravity position, target foot position / posture, target arm posture (target antiphase arm)
- the instantaneous value of a variable representing the motion (including the swing angle) and the like (this is called the motion variable) and the instantaneous value of the target ZMP are input to the inverse dynamics full model (inverse full model) 201 .
- Balance the motion represented by the input motion variables ie, The floor reaction force horizontal component and the floor reaction force moment about the target ZMP (horizontal component and vertical component) Force Calculated by the inverse full model 201 calculation process.
- the floor reaction force moment around the target ZMP calculated by the inverse full model 210 has a meaning as an error of the simplified model gait.
- the horizontal component of the floor reaction force and the horizontal component of the floor reaction force and the vertical component of the floor reaction force moment obtained by the inverse full model 201 are referred to as the “full model floor reaction force horizontal component” and the “full model floor reaction force component,” respectively.
- the full model floor reaction force horizontal component is often abbreviated as Ffull
- the full model floor reaction force moment horizontal component is Mfulbcy
- the full model floor reaction force moment vertical component is Mfullz.
- the target overall center of gravity vertical position is input, and the desired floor reaction force vertical component is obtained from the second derivative of the target overall center of gravity vertical position.
- the target floor reaction force vertical component may be input to the full model for reasons such as reducing the number of calculations even if it is redundant.
- the perturbation model is composed of a perturbation model for body horizontal position correction 202, a perturbation model for body posture inclination angle correction 203, and a perturbation model for antiphase arm swing angle correction 2 31.
- the perturbation model may be a single model as shown in Fig. 12 without being separated into three models in this way.
- the perturbation model for body posture inclination correction 203 is shown in the figure. Abbreviated. '
- the body horizontal position correction perturbation model 202 represents the relationship between the perturbation of the floor reaction force and the perturbation of the body horizontal position in the body translation mode.
- the body horizontal position correction perturbation model 202 inputs a correction amount of the desired floor reaction force moment, and calculates a correction amount of the target body horizontal position that is dynamically balanced with the input. This input (correction amount of the desired floor reaction force moment) is called the perturbation model moment Mp for body horizontal position correction. Also, the output of the perturbation model 202 for body horizontal position correction (correction amount of the target body horizontal position) is called the correction perturbation model body horizontal position Xc.
- the floor reaction force horizontal component generated by the body horizontal position correction perturbation model 202 is referred to as a body horizontal position correction perturbation model floor reaction horizontal component Fp.
- the perturbation model for body horizontal position correction 202 is represented by an inverted pendulum consisting of a fulcrum, an inverted pendulum mass (upper mass), and a telescopic support rod connecting these.
- the horizontal position of the fulcrum coincides with the horizontal position of the origin of the support leg coordinate system of the current time's gait, and the fulcrum position is set so that the vertical position of the fulcrum coincides with the vertical position of the target ZMP.
- the mass mb of the inverted pendulum mass is the same as the mass of the upper mass of the simplified model (3 mass points + flywheel model) shown in Fig. 12.
- the vertical position Zc of the inverted pendulum mass point is the same as the vertical position Zb of the upper body mass position of the simplified model shown in Fig. 12 corresponding to the simplified gait.
- the body horizontal position correction perturbation model 202 represents the relationship between the floor reaction force moment perturbation A My and the body mass point horizontal position perturbation A Xb in the simplified model.
- Body translational mode floor reaction force ratio h which is the ratio of ⁇ Mp and AFp generated by the body horizontal acceleration, is the upper body of the right side of equation al 2 Since it is the ratio of the term generated by the horizontal acceleration (that is, the second term) and the equation al3, the following equation is obtained.
- h (Zb-Zzmp) Equation a 1 4 That is, the body translational mode floor reaction force ratio h is equivalent to the height of the simplified model's upper body mass point (inverted pendulum mass point) from the fulcrum.
- Fbz Fz- msup * ( g + d2Zsup / dt2)-mswg * (g + d2Zswg / dt2)
- Equation a 1 7 the body floor reaction force vertical component is the floor reaction force vertical component Fz and the gravity acting on both legs mass points of the simplified model (3 mass points + flywheel model) shown in Fig. 12. It is equal to the sum of the translational force and the vertical component of the resultant force of the inertia force.
Landscapes
- Engineering & Computer Science (AREA)
- Mechanical Engineering (AREA)
- Robotics (AREA)
- Chemical & Material Sciences (AREA)
- Combustion & Propulsion (AREA)
- Transportation (AREA)
- Manipulator (AREA)
- Control Of Position Or Direction (AREA)
- Numerical Control (AREA)
Abstract
Description
Claims
Priority Applications (6)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
DE602004031526T DE602004031526D1 (de) | 2003-06-27 | 2004-06-28 | Steuervorrichtung für mobilen schreitroboter |
US10/562,168 US8005573B2 (en) | 2003-06-27 | 2004-06-28 | Control device for legged mobile robot |
EP04746941A EP1642687B9 (en) | 2003-06-27 | 2004-06-28 | Control device for legged mobile robot |
JP2005511135A JP4155993B2 (ja) | 2003-06-27 | 2004-06-28 | 脚式移動ロボットの制御装置 |
KR1020057019866A KR101083417B1 (ko) | 2003-06-27 | 2004-06-28 | 다리식 이동 로봇의 제어장치 |
US12/603,116 US7860611B2 (en) | 2003-06-27 | 2009-10-21 | Control device for legged mobile robot |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2003-185930 | 2003-06-27 | ||
JP2003185930 | 2003-06-27 |
Related Child Applications (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US10562168 A-371-Of-International | 2004-06-28 | ||
US12/603,116 Division US7860611B2 (en) | 2003-06-27 | 2009-10-21 | Control device for legged mobile robot |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2005000534A1 true WO2005000534A1 (ja) | 2005-01-06 |
Family
ID=33549677
Family Applications (3)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2004/009472 WO2005000534A1 (ja) | 2003-06-27 | 2004-06-28 | 脚式移動ロボットの制御装置 |
PCT/JP2004/009476 WO2005000535A1 (ja) | 2003-06-27 | 2004-06-28 | 脚式移動ロボットの歩容生成装置および脚式移動ロボットの制御装置 |
PCT/JP2004/009477 WO2005000536A1 (ja) | 2003-06-27 | 2004-06-28 | 脚式移動ロボットの制御装置 |
Family Applications After (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2004/009476 WO2005000535A1 (ja) | 2003-06-27 | 2004-06-28 | 脚式移動ロボットの歩容生成装置および脚式移動ロボットの制御装置 |
PCT/JP2004/009477 WO2005000536A1 (ja) | 2003-06-27 | 2004-06-28 | 脚式移動ロボットの制御装置 |
Country Status (6)
Country | Link |
---|---|
US (4) | US7379789B2 (ja) |
EP (10) | EP1642689B1 (ja) |
JP (3) | JP4126061B2 (ja) |
KR (3) | KR101083414B1 (ja) |
DE (2) | DE602004031425D1 (ja) |
WO (3) | WO2005000534A1 (ja) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2006247769A (ja) * | 2005-03-09 | 2006-09-21 | Toyota Motor Corp | 脚式ロボットとその動作制御方法 |
WO2008099907A1 (ja) | 2007-02-16 | 2008-08-21 | Ono Pharmaceutical Co., Ltd. | 尿排出障害治療剤 |
US7603199B2 (en) | 2003-11-27 | 2009-10-13 | Honda Motor Co., Ltd. | Control device for mobile body |
CN111872941A (zh) * | 2020-08-06 | 2020-11-03 | 深圳市优必选科技股份有限公司 | 平衡控制方法、装置、仿人机器人及可读存储介质 |
Families Citing this family (103)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7774177B2 (en) * | 2001-06-29 | 2010-08-10 | Honda Motor Co., Ltd. | Exoskeleton controller for a human-exoskeleton system |
US7650204B2 (en) * | 2001-06-29 | 2010-01-19 | Honda Motor Co., Ltd. | Active control of an ankle-foot orthosis |
US7319917B2 (en) * | 2001-12-28 | 2008-01-15 | Honda Giken Kogyo Kabushiki Kaisha | Gait generation device for legged mobile robot |
US7319918B2 (en) * | 2001-12-28 | 2008-01-15 | Honda Giken Kogyo Kabushiki Kaisha | Gait generation device for legged mobile robot |
DE602004031425D1 (de) * | 2003-06-27 | 2011-03-31 | Honda Motor Co Ltd | Steuerung für mobilen roboter mit beinen |
DE602004029256D1 (de) * | 2003-06-27 | 2010-11-04 | Honda Motor Co Ltd | Gangerzeugungsvorrichtung für beweglichen roboter mit beinen |
US7400108B2 (en) * | 2004-04-15 | 2008-07-15 | University Of Utah Research Foundation | System and method for controlling modular robots |
JP4485279B2 (ja) * | 2004-08-02 | 2010-06-16 | 本田技研工業株式会社 | 脚式移動ロボットの歩容生成装置および制御装置 |
JP2006068872A (ja) * | 2004-09-03 | 2006-03-16 | Honda Motor Co Ltd | 脚式移動ロボット |
US8082062B2 (en) * | 2005-06-10 | 2011-12-20 | Honda Motor Co., Ltd. | Regenerative actuation in motion control |
JP4812426B2 (ja) * | 2005-12-27 | 2011-11-09 | 富士通株式会社 | ロボット制御装置 |
JP4908020B2 (ja) * | 2006-03-02 | 2012-04-04 | 本田技研工業株式会社 | ハンド制御システム |
JP2008073830A (ja) * | 2006-09-25 | 2008-04-03 | Fanuc Ltd | ロボット制御装置 |
WO2008086629A1 (en) | 2007-01-19 | 2008-07-24 | Victhom Human Bionics Inc. | Reactive layer control system for prosthetic and orthotic devices |
WO2009040885A1 (ja) * | 2007-09-25 | 2009-04-02 | Fujitsu Limited | ロボット制御装置、ロボット制御方法およびロボット制御プログラム |
KR101182620B1 (ko) * | 2007-10-23 | 2012-09-14 | 혼다 기켄 고교 가부시키가이샤 | 2 족 보행 로봇 |
JP4950018B2 (ja) * | 2007-12-25 | 2012-06-13 | アルプス電気株式会社 | 座標入力装置 |
JP5104355B2 (ja) * | 2008-02-01 | 2012-12-19 | 富士通株式会社 | ロボット制御装置、ロボット制御方法およびロボット制御プログラム |
FR2926971B1 (fr) * | 2008-02-04 | 2011-02-25 | Commissariat Energie Atomique | Dispositif d'analyse de la marche |
KR101464124B1 (ko) * | 2008-06-04 | 2014-11-20 | 삼성전자주식회사 | 로봇 및 그 보행제어방법 |
US20100042357A1 (en) * | 2008-08-15 | 2010-02-18 | Oceaneering International, Inc. | Manipulator Position Sensor System |
FR2947236B1 (fr) * | 2009-06-30 | 2011-07-15 | Aldebaran Robotics | Procede pour controler la marche d'un robot mobile et robot mettant en oeuvre le procede |
CA2769975A1 (en) * | 2009-07-01 | 2011-01-06 | Rex Bionics Limited | Control system for a mobility aid |
JP5232120B2 (ja) * | 2009-10-01 | 2013-07-10 | 本田技研工業株式会社 | 移動体の制御装置 |
JP5232124B2 (ja) * | 2009-10-28 | 2013-07-10 | 本田技研工業株式会社 | 脚式移動ロボットの制御装置 |
KR20110082712A (ko) * | 2010-01-12 | 2011-07-20 | 삼성전자주식회사 | 로봇 및 그 제어방법 |
KR101687628B1 (ko) * | 2010-01-12 | 2016-12-21 | 삼성전자주식회사 | 로봇의 보행 제어 장치 및 그 제어 방법 |
FR2958152A1 (fr) | 2010-03-31 | 2011-10-07 | Benjamin Penot | Systeme d'articulation complexe pour orthese, prothese, robotique, exosquelette |
KR101200191B1 (ko) * | 2010-07-14 | 2012-11-13 | 서울대학교산학협력단 | 데이터 기반 바이페드 제어 장치 및 방법 |
KR101766755B1 (ko) * | 2010-08-17 | 2017-08-10 | 삼성전자주식회사 | 보행 로봇 및 그 제어방법 |
KR20120024098A (ko) * | 2010-09-06 | 2012-03-14 | 삼성전자주식회사 | 보행 로봇 및 그 제어방법 |
KR101760883B1 (ko) * | 2010-09-09 | 2017-08-04 | 삼성전자주식회사 | 로봇 및 그 제어방법 |
US8467948B2 (en) * | 2010-09-29 | 2013-06-18 | Honda Motor Co., Ltd. | Omnidirectional moving body operation system and omnidirectional moving body operation method |
WO2012047721A1 (en) | 2010-09-29 | 2012-04-12 | össur hf | Prosthetic and orthotic devices and methods and systems for controlling the same |
KR20120069924A (ko) * | 2010-12-21 | 2012-06-29 | 삼성전자주식회사 | 보행 로봇 및 그 제어 방법 |
KR20120071599A (ko) * | 2010-12-23 | 2012-07-03 | 삼성전자주식회사 | 보행 로봇 및 그 제어 방법 |
US8880221B2 (en) | 2011-03-21 | 2014-11-04 | Honda Motor Co., Ltd. | Damage reduction control for humanoid robot fall |
US9060884B2 (en) | 2011-05-03 | 2015-06-23 | Victhom Human Bionics Inc. | Impedance simulating motion controller for orthotic and prosthetic applications |
US8504208B2 (en) * | 2011-05-25 | 2013-08-06 | Honda Motor Co., Ltd. | Mobile object controller and floor surface estimator |
FR2976700B1 (fr) * | 2011-06-17 | 2013-07-12 | Inst Nat Rech Inf Automat | Procede de generation d'ordres de commande de coordination d'organes de deplacement d'une plateforme animee et generateur correspondant. |
JP5752500B2 (ja) * | 2011-06-27 | 2015-07-22 | 本田技研工業株式会社 | 軌道生成システム |
JP5724855B2 (ja) * | 2011-12-02 | 2015-05-27 | トヨタ自動車株式会社 | 倒立移動体及び角速度センサの出力値補正方法 |
KR20130063230A (ko) * | 2011-12-06 | 2013-06-14 | 삼성전자주식회사 | 보행 로봇 및 그 제어 방법 |
JP5945419B2 (ja) * | 2012-01-10 | 2016-07-05 | 本田技研工業株式会社 | 脚式移動ロボットの脚体運動軌道生成装置。 |
US9605952B2 (en) * | 2012-03-08 | 2017-03-28 | Quality Manufacturing Inc. | Touch sensitive robotic gripper |
US9156159B2 (en) * | 2012-03-08 | 2015-10-13 | Disney Enterprises Inc. | Robot cyclic locomotion using a dynamic object |
US9044346B2 (en) | 2012-03-29 | 2015-06-02 | össur hf | Powered prosthetic hip joint |
JP6110636B2 (ja) * | 2012-08-09 | 2017-04-05 | 日本電産サンキョー株式会社 | 産業用ロボット |
JP6065502B2 (ja) * | 2012-09-28 | 2017-01-25 | 三菱商事株式会社 | フラーレン誘導体及びその製造方法 |
KR101371756B1 (ko) * | 2012-12-17 | 2014-03-12 | 현대자동차(주) | 로봇의 보행제어방법 |
JP5642214B2 (ja) * | 2013-02-15 | 2014-12-17 | 株式会社神戸製鋼所 | 多関節ロボットの弾性変形補償制御装置 |
JP5888309B2 (ja) * | 2013-10-31 | 2016-03-22 | カシオ計算機株式会社 | トレーニング支援装置およびシステム、フォーム解析装置および方法、ならびにプログラム |
CN105934724B (zh) * | 2014-01-23 | 2018-12-21 | 三菱电机株式会社 | 电动机控制装置 |
US9292786B2 (en) * | 2014-02-03 | 2016-03-22 | Disney Enterprises, Inc. | Universal balancing controller for lateral stabilization of bipedal robots in dynamic unstable environments |
JP5897644B2 (ja) * | 2014-06-02 | 2016-03-30 | 株式会社神戸製鋼所 | ロボットの制御装置 |
US9849926B2 (en) * | 2014-07-23 | 2017-12-26 | Boston Dynamics, Inc. | Predictively adjustable hydraulic pressure rails |
US9308648B2 (en) * | 2014-07-24 | 2016-04-12 | Google Inc. | Systems and methods for robotic self-right |
US9499219B1 (en) | 2014-08-25 | 2016-11-22 | Google Inc. | Touch-down sensing for robotic devices |
US9387588B1 (en) * | 2014-08-25 | 2016-07-12 | Google Inc. | Handling gait disturbances with asynchronous timing |
US10081098B1 (en) | 2014-08-25 | 2018-09-25 | Boston Dynamics, Inc. | Generalized coordinate surrogates for integrated estimation and control |
US9387896B1 (en) | 2014-08-25 | 2016-07-12 | Google Inc. | Slip avoidance |
US9517561B2 (en) | 2014-08-25 | 2016-12-13 | Google Inc. | Natural pitch and roll |
US9618937B1 (en) | 2014-08-25 | 2017-04-11 | Google Inc. | Slip detection using robotic limbs |
US9623556B1 (en) * | 2014-09-03 | 2017-04-18 | X Development Llc | Robotic sole joint |
US9352470B1 (en) * | 2014-11-11 | 2016-05-31 | Google Inc. | Yaw slip handling in a robotic device |
US9446518B1 (en) * | 2014-11-11 | 2016-09-20 | Google Inc. | Leg collision avoidance in a robotic device |
US9519289B2 (en) * | 2014-11-26 | 2016-12-13 | Irobot Corporation | Systems and methods for performing simultaneous localization and mapping using machine vision systems |
US9751210B2 (en) | 2014-11-26 | 2017-09-05 | Irobot Corporation | Systems and methods for performing occlusion detection |
US9440353B1 (en) | 2014-12-29 | 2016-09-13 | Google Inc. | Offline determination of robot behavior |
US9499218B1 (en) | 2014-12-30 | 2016-11-22 | Google Inc. | Mechanically-timed footsteps for a robotic device |
WO2016159346A1 (ja) * | 2015-04-02 | 2016-10-06 | 国立大学法人大阪大学 | 脚型機構体、歩行ロボット、姿勢制御方法及びプログラム |
US9594377B1 (en) * | 2015-05-12 | 2017-03-14 | Google Inc. | Auto-height swing adjustment |
US9561592B1 (en) * | 2015-05-15 | 2017-02-07 | Google Inc. | Ground plane compensation for legged robots |
JP6450279B2 (ja) * | 2015-08-05 | 2019-01-09 | 本田技研工業株式会社 | 移動ロボットの目標zmp軌道の生成装置 |
US9586316B1 (en) | 2015-09-15 | 2017-03-07 | Google Inc. | Determination of robotic step path |
JP6498597B2 (ja) * | 2015-12-14 | 2019-04-10 | 本田技研工業株式会社 | 移動ロボットの制御装置 |
US9778132B1 (en) * | 2015-12-16 | 2017-10-03 | X Development Llc | Methods and systems for force sensor calibration |
JP6483014B2 (ja) * | 2015-12-25 | 2019-03-13 | 本田技研工業株式会社 | 移動ロボットの制御装置 |
US9868210B1 (en) * | 2015-12-30 | 2018-01-16 | Google Inc. | Methods and systems for planning a body position of a robotic device |
US9925667B1 (en) | 2016-01-25 | 2018-03-27 | Boston Dynamics, Inc. | Continuous slip recovery |
US9789919B1 (en) | 2016-03-22 | 2017-10-17 | Google Inc. | Mitigating sensor noise in legged robots |
US9987745B1 (en) | 2016-04-01 | 2018-06-05 | Boston Dynamics, Inc. | Execution of robotic tasks |
US10821614B2 (en) | 2016-11-11 | 2020-11-03 | Sarcos Corp. | Clutched joint modules having a quasi-passive elastic actuator for a robotic assembly |
US10828767B2 (en) | 2016-11-11 | 2020-11-10 | Sarcos Corp. | Tunable actuator joint modules having energy recovering quasi-passive elastic actuators with internal valve arrangements |
CN109693234B (zh) * | 2017-10-20 | 2021-08-27 | 深圳市优必选科技有限公司 | 机器人跌倒预测方法、装置、终端设备及计算机存储介质 |
US10780578B2 (en) * | 2018-02-22 | 2020-09-22 | Boston Dynamics, Inc. | Reaching mobile robots |
US10946518B2 (en) * | 2018-07-24 | 2021-03-16 | Invia Robotics, Inc. | Spatiotemporal controller for controlling robot operation |
WO2020107279A1 (zh) * | 2018-11-28 | 2020-06-04 | 深圳市优必选科技有限公司 | 双足机器人及其移动方法、装置和存储介质 |
US11241801B2 (en) | 2018-12-31 | 2022-02-08 | Sarcos Corp. | Robotic end effector with dorsally supported actuation mechanism |
CN110597267B (zh) * | 2019-09-27 | 2023-01-10 | 长安大学 | 一种足式机器人的局部最优落足点选取方法 |
US11292126B2 (en) * | 2019-10-17 | 2022-04-05 | Disney Enterprises, Inc. | Robots with robust bipedal locomotion supported with non-conventional physics |
CN111651900B (zh) * | 2020-06-29 | 2023-07-04 | 中国有色金属工业昆明勘察设计研究院有限公司 | 一种昔格达地层岩质边坡稳定性计算的拟动力上限法 |
US11833676B2 (en) | 2020-12-07 | 2023-12-05 | Sarcos Corp. | Combining sensor output data to prevent unsafe operation of an exoskeleton |
CN113246123B (zh) * | 2021-04-30 | 2022-10-18 | 深圳市优必选科技股份有限公司 | 机器人控制方法、装置、计算机可读存储介质及机器人 |
DE102021112485B3 (de) | 2021-05-12 | 2022-08-11 | Deutsches Zentrum für Luft- und Raumfahrt e.V. | Verfahren zur Balancierung eines Roboters, Verfahren zur Ganzkörpersteuerung eines Roboters, Regler sowie Roboter |
CN113359800B (zh) * | 2021-06-30 | 2024-08-09 | 深圳市优必选科技股份有限公司 | 机器人行走控制方法、装置、机器人控制设备及存储介质 |
CN113377113B (zh) * | 2021-07-02 | 2022-01-04 | 北方工业大学 | 足式机器人足端轨迹规划方法、系统及控制方法、系统 |
US11738452B1 (en) * | 2022-07-29 | 2023-08-29 | Sarcos Corp. | Sole with various compliant regions for robots |
US11826907B1 (en) | 2022-08-17 | 2023-11-28 | Sarcos Corp. | Robotic joint system with length adapter |
US11717956B1 (en) | 2022-08-29 | 2023-08-08 | Sarcos Corp. | Robotic joint system with integrated safety |
US11924023B1 (en) | 2022-11-17 | 2024-03-05 | Sarcos Corp. | Systems and methods for redundant network communication in a robot |
US11897132B1 (en) | 2022-11-17 | 2024-02-13 | Sarcos Corp. | Systems and methods for redundant network communication in a robot |
CN117963038B (zh) * | 2024-04-02 | 2024-05-28 | 吉林大学 | 一种具有缓冲与抗沉陷功能的刚柔耦合仿生机械足 |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH0213593A (ja) | 1988-06-30 | 1990-01-17 | Mitsubishi Electric Corp | 車椅子乗用踏段付エスカレータ |
EP0572285A1 (en) | 1992-05-29 | 1993-12-01 | Honda Giken Kogyo Kabushiki Kaisha | Attitude stabilization control system for a mobile robot, especially a legged mobile robot |
WO2002040224A1 (fr) * | 2000-11-17 | 2002-05-23 | Honda Giken Kogyo Kabushiki Kaisha | Dispositif generateur d'un modele de demarche pour robot mobile pourvu de jambes |
JP2002326173A (ja) * | 2001-04-27 | 2002-11-12 | Honda Motor Co Ltd | 脚式移動ロボットの動作生成装置 |
Family Cites Families (30)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPS60221288A (ja) * | 1984-04-13 | 1985-11-05 | 株式会社 富士電機総合研究所 | 圧覚認識制御装置 |
JPH0213596A (ja) | 1988-06-28 | 1990-01-17 | Aichi Steel Works Ltd | クレーンのトング装置 |
JP2581175B2 (ja) | 1988-06-30 | 1997-02-12 | 三菱電機株式会社 | 車椅子乗用踏段付きエスカレータ |
JPH0213784A (ja) | 1988-06-30 | 1990-01-18 | Kaneko Agricult Mach Co Ltd | 乾燥施設における試料穀物の乾燥方法およびその装置 |
JP2548799B2 (ja) | 1989-05-29 | 1996-10-30 | 日鋼特機株式会社 | 管の縮径成形装置 |
JPH034355A (ja) | 1989-05-31 | 1991-01-10 | Nec Corp | データ伝送方式 |
US5355064A (en) * | 1992-03-04 | 1994-10-11 | Honda Giken Kogyo Kabushiki Kaisha | Control system for legged mobile robot |
JP3148827B2 (ja) | 1992-04-30 | 2001-03-26 | 本田技研工業株式会社 | 脚式移動ロボットの歩行制御装置 |
JP3233450B2 (ja) | 1992-05-22 | 2001-11-26 | 本田技研工業株式会社 | 指定時刻到達関数発生器 |
JP3330710B2 (ja) * | 1993-12-30 | 2002-09-30 | 本田技研工業株式会社 | 移動ロボットの位置検知および制御装置 |
US5644204A (en) * | 1994-11-03 | 1997-07-01 | Nagle; John | Anti-slip control for a legged robot and realisitc simulation of a legged creature |
JP3658147B2 (ja) | 1996-07-25 | 2005-06-08 | 本田技研工業株式会社 | 脚式移動ロボットの歩容生成装置 |
JP3663034B2 (ja) | 1996-07-25 | 2005-06-22 | 本田技研工業株式会社 | 脚式移動ロボットの歩容生成装置 |
US6243623B1 (en) * | 1997-01-31 | 2001-06-05 | Honda Giken Kogyo Kabushiki Kaisha | Leg type mobile robot control apparatus |
JP3672406B2 (ja) * | 1997-01-31 | 2005-07-20 | 本田技研工業株式会社 | 脚式移動ロボットの歩容生成装置 |
JP3629133B2 (ja) | 1997-01-31 | 2005-03-16 | 本田技研工業株式会社 | 脚式移動ロボットの制御装置 |
JP3629143B2 (ja) | 1998-04-20 | 2005-03-16 | 本田技研工業株式会社 | 脚式移動ロボットの制御装置 |
EP1120203B1 (en) | 1998-04-20 | 2011-01-19 | Honda Giken Kogyo Kabushiki Kaisha | Controller for legged mobile robot |
JP2000352011A (ja) | 1999-06-11 | 2000-12-19 | Gakunan Koki Kk | 視覚障害者用警告形道路ラインの施工方法 |
JP2001133621A (ja) | 1999-11-08 | 2001-05-18 | Sharp Corp | カラーフィルターの製造方法および製造装置 |
JP4480843B2 (ja) | 2000-04-03 | 2010-06-16 | ソニー株式会社 | 脚式移動ロボット及びその制御方法、並びに、脚式移動ロボット用相対移動測定センサ |
JP3726009B2 (ja) * | 2000-05-19 | 2005-12-14 | 本田技研工業株式会社 | 脚式移動ロボットの床形状推定装置 |
US7135003B2 (en) * | 2001-06-29 | 2006-11-14 | Honda Giken Kogyo Kabushiki Kaisha | Feedback estimation of joint forces and joint moments |
US7319918B2 (en) * | 2001-12-28 | 2008-01-15 | Honda Giken Kogyo Kabushiki Kaisha | Gait generation device for legged mobile robot |
US7319917B2 (en) * | 2001-12-28 | 2008-01-15 | Honda Giken Kogyo Kabushiki Kaisha | Gait generation device for legged mobile robot |
EP1467265B8 (en) | 2001-12-28 | 2009-07-08 | Honda Giken Kogyo Kabushiki Kaisha | Control system |
KR100924505B1 (ko) | 2002-01-18 | 2009-11-02 | 혼다 기켄 고교 가부시키가이샤 | 다리식 이동로봇의 제어장치 |
WO2003090979A1 (fr) * | 2002-04-26 | 2003-11-06 | Honda Giken Kogyo Kabushiki Kaisha | Systeme permettant d'estimer l'attitude d'un robot mobile monte sur des jambes |
WO2003090982A1 (fr) * | 2002-04-26 | 2003-11-06 | Honda Giken Kogyo Kabushiki Kaisha | Dispositif de commande et dispositif de determination de pas pour robot mobile sur jambes |
DE602004031425D1 (de) * | 2003-06-27 | 2011-03-31 | Honda Motor Co Ltd | Steuerung für mobilen roboter mit beinen |
-
2004
- 2004-06-28 DE DE602004031425T patent/DE602004031425D1/de not_active Expired - Lifetime
- 2004-06-28 EP EP04746946A patent/EP1642689B1/en not_active Expired - Lifetime
- 2004-06-28 US US10/561,988 patent/US7379789B2/en active Active
- 2004-06-28 US US10/562,168 patent/US8005573B2/en not_active Expired - Fee Related
- 2004-06-28 EP EP04746941A patent/EP1642687B9/en not_active Expired - Lifetime
- 2004-06-28 WO PCT/JP2004/009472 patent/WO2005000534A1/ja active Application Filing
- 2004-06-28 EP EP11003058A patent/EP2353794B1/en not_active Expired - Lifetime
- 2004-06-28 EP EP10003333A patent/EP2210713B1/en not_active Expired - Lifetime
- 2004-06-28 KR KR1020057020794A patent/KR101083414B1/ko active IP Right Grant
- 2004-06-28 EP EP10003336A patent/EP2206582B1/en not_active Expired - Lifetime
- 2004-06-28 JP JP2005511138A patent/JP4126061B2/ja not_active Expired - Fee Related
- 2004-06-28 WO PCT/JP2004/009476 patent/WO2005000535A1/ja active Application Filing
- 2004-06-28 EP EP04746945A patent/EP1642688B1/en not_active Expired - Lifetime
- 2004-06-28 WO PCT/JP2004/009477 patent/WO2005000536A1/ja active Application Filing
- 2004-06-28 EP EP11003059A patent/EP2353795B1/en not_active Expired - Lifetime
- 2004-06-28 JP JP2005511139A patent/JP4181175B2/ja not_active Expired - Fee Related
- 2004-06-28 KR KR1020057019866A patent/KR101083417B1/ko active IP Right Grant
- 2004-06-28 JP JP2005511135A patent/JP4155993B2/ja not_active Expired - Fee Related
- 2004-06-28 EP EP10003335A patent/EP2208582B1/en not_active Expired - Lifetime
- 2004-06-28 EP EP10003337A patent/EP2206583B1/en not_active Expired - Lifetime
- 2004-06-28 EP EP10003338A patent/EP2206584B1/en not_active Expired - Lifetime
- 2004-06-28 KR KR1020057019907A patent/KR101083412B1/ko active IP Right Grant
- 2004-06-28 DE DE602004031526T patent/DE602004031526D1/de not_active Expired - Lifetime
- 2004-06-28 US US10/562,327 patent/US7603234B2/en not_active Expired - Fee Related
-
2009
- 2009-10-21 US US12/603,116 patent/US7860611B2/en not_active Expired - Fee Related
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH0213593A (ja) | 1988-06-30 | 1990-01-17 | Mitsubishi Electric Corp | 車椅子乗用踏段付エスカレータ |
EP0572285A1 (en) | 1992-05-29 | 1993-12-01 | Honda Giken Kogyo Kabushiki Kaisha | Attitude stabilization control system for a mobile robot, especially a legged mobile robot |
JPH05337849A (ja) | 1992-05-29 | 1993-12-21 | Honda Motor Co Ltd | 脚式移動ロボットの姿勢安定化制御装置 |
WO2002040224A1 (fr) * | 2000-11-17 | 2002-05-23 | Honda Giken Kogyo Kabushiki Kaisha | Dispositif generateur d'un modele de demarche pour robot mobile pourvu de jambes |
JP2002326173A (ja) * | 2001-04-27 | 2002-11-12 | Honda Motor Co Ltd | 脚式移動ロボットの動作生成装置 |
Non-Patent Citations (1)
Title |
---|
See also references of EP1642687A4 * |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7603199B2 (en) | 2003-11-27 | 2009-10-13 | Honda Motor Co., Ltd. | Control device for mobile body |
US7606634B2 (en) | 2003-11-27 | 2009-10-20 | Honda Motor Co., Ltd. | Control device for mobile body |
JP2006247769A (ja) * | 2005-03-09 | 2006-09-21 | Toyota Motor Corp | 脚式ロボットとその動作制御方法 |
JP4492395B2 (ja) * | 2005-03-09 | 2010-06-30 | トヨタ自動車株式会社 | 脚式ロボットとその動作制御方法 |
WO2008099907A1 (ja) | 2007-02-16 | 2008-08-21 | Ono Pharmaceutical Co., Ltd. | 尿排出障害治療剤 |
CN111872941A (zh) * | 2020-08-06 | 2020-11-03 | 深圳市优必选科技股份有限公司 | 平衡控制方法、装置、仿人机器人及可读存储介质 |
CN111872941B (zh) * | 2020-08-06 | 2021-09-07 | 深圳市优必选科技股份有限公司 | 平衡控制方法、装置、仿人机器人及可读存储介质 |
Also Published As
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2005000534A1 (ja) | 脚式移動ロボットの制御装置 | |
JP3679105B2 (ja) | 脚式移動ロボットの歩容生成装置 | |
JP4246638B2 (ja) | 脚式移動ロボットの制御装置 | |
JP3672101B2 (ja) | 脚式移動ロボットの歩容生成装置 | |
WO2005000533A1 (ja) | 脚式移動ロボットの歩容生成装置 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): AE AG AL AM AT AU AZ BA BB BG BR BW BY BZ CA CH CN CO CR CU CZ DE DK DM DZ EC EE EG ES FI GB GD GE GH GM HR HU ID IL IN IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MA MD MG MK MN MW MX MZ NA NI NO NZ OM PG PH PL PT RO RU SC SD SE SG SK SL SY TJ TM TN TR TT TZ UA UG US UZ VC VN YU ZA ZM ZW |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): BW GH GM KE LS MW MZ NA SD SL SZ TZ UG ZM ZW AM AZ BY KG KZ MD RU TJ TM AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IT LU MC NL PL PT RO SE SI SK TR BF BJ CF CG CI CM GA GN GQ GW ML MR NE SN TD TG |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
WWE | Wipo information: entry into national phase |
Ref document number: 2005511135 Country of ref document: JP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1020057019866 Country of ref document: KR |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2006247799 Country of ref document: US Ref document number: 10562168 Country of ref document: US |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2004746941 Country of ref document: EP |
|
WWP | Wipo information: published in national office |
Ref document number: 1020057019866 Country of ref document: KR |
|
WWP | Wipo information: published in national office |
Ref document number: 2004746941 Country of ref document: EP |
|
WWP | Wipo information: published in national office |
Ref document number: 10562168 Country of ref document: US |