JP4905041B2 - Robot controller - Google Patents

Description

  The present invention relates to a robot control device that controls walking of a robot, and more particularly to a robot control device that can detect a change in a walking environment and make the robot respond to the environmental change.

  In recent years, the walking of humanoid robots, especially humanoid robots, has attracted the attention of many researchers. Most of the research on the walking of this humanoid robot uses the ZMP (Zero Moment Point) norm. This ZMP norm controls to keep the ZMP inside the support polygon. In this approach, the humanoid robot and the surrounding environment of the robot are accurately modeled and the differential equation is solved. However, this modeling becomes difficult when there are unknown elements. Furthermore, since it takes time to solve the differential equation, real-time control becomes difficult.

  One approach is to not use the ZMP norm. For example, there is a conventional technique that uses the periodic motion of a movable part of a robot to adjust the phase of the periodic motion so that the posture of the robot is stabilized (see Patent Document 1). Here, the movable part is a leg or arm of the robot.

  Japanese Patent Application No. 2005-37659 describes a technique that allows a humanoid robot to efficiently perform various motions while eliminating the need for modeling the humanoid robot and the surrounding environment of the robot. Yes.

Japanese Patent Laid-Open No. 2005-96068

  However, in this conventional technology, when there is a change in the walking environment, such as when an obstacle is placed on the walking surface or there is a step on the walking surface, the humanoid robot cannot continue the periodic motion. Therefore, there is a problem that the humanoid robot cannot be placed under control.

  The present invention has been made to solve the above-described problems caused by the prior art, and it is an object of the present invention to provide a robot control device capable of detecting a change in a walking environment and adapting the robot to the environmental change. To do.

In order to solve the above-described problems and achieve the object, the present invention is a robot control apparatus that controls walking of a robot, and includes a plurality of different time points including at least a reference posture in which the robot stands alone without falling down. Control information is generated based on the posture of the robot, and a walking motion control means for controlling the robot to perform a predetermined walking motion, and a change in the walking environment that makes the continuation of the predetermined walking motion impossible. Reflection control means for detecting and correcting control of walking motion by the walking motion control means in response to the detected change.

According to the present invention, control information is generated based on postures at a plurality of different times including at least a reference posture where the robot stands alone without falling, and control is performed so that the robot performs a predetermined walking motion. Since it is configured to detect a change in the walking environment that makes it impossible to continue the predetermined walking motion, and to correct the control of the walking motion in response to the detected change, the control corresponding to the change in the walking environment It can be performed.

In the present invention, when the reflection control means detects the presence of an obstacle placed at the place where the robot's leg lands, the reflection control means performs a rolling operation to the support leg and an operation to return the swinging leg to the position before the swing. The walking motion control means is instructed to carry out.

According to the present invention, when the presence of an obstacle placed at the place where the robot leg lands is detected, control is performed so as to perform a rolling operation to the support leg and an operation to return the swinging leg to the position before the swing. Therefore, control corresponding to the obstacle can be performed.

In the present invention, when the reflection control means detects that the walking surface on which the robot walks is lowered, the reflection control means instructs the walking action control means to extend the walking action time and continue the landing action. Features.

According to the present invention, when it is detected that the walking surface on which the robot walks is low, the walking operation time is lengthened and the landing operation is controlled so that the control corresponding to the decrease in the walking surface is performed. It can be carried out.

Further, in the present invention, when the reflection control means detects that the walking surface on which the robot walks is high, the reflection control means instructs the walking movement control means to lengthen the walking movement time and raise the landing leg. It is characterized by.

According to the present invention, when it is detected that the walking surface on which the robot walks is high, the walking operation time is lengthened and control is performed to raise the landing leg. Control can be performed.

In the present invention, the reflection control means may be configured to reduce the gain of the gyro feedback control to a predetermined value and rotate the ankle and waist when the weight of the load held by the robot suddenly changes. The operation control means is instructed.

According to the present invention, when the weight of the load held by the robot suddenly changes, the gain of the gyro feedback control is reduced to a predetermined value and the ankle and the waist are controlled to rotate. It is possible to perform control corresponding to the fall of the

  According to the present invention, since the control corresponding to the change of the walking environment is performed, there is an effect that the robot can be adapted to the change of the walking environment.

  In addition, according to the present invention, since control corresponding to an obstacle is performed, there is an effect that the robot can correspond to the obstacle.

  Further, according to the present invention, since control corresponding to a decrease in the walking surface is performed, there is an effect that the robot can be adapted to the decrease in the walking surface.

  Further, according to the present invention, the control corresponding to the up step on the walking surface is performed, so that the robot can be made to correspond to the up step on the walking surface.

  Further, according to the present invention, since control corresponding to the fall of the load is performed, there is an effect that the robot can respond to the fall of the load.

  Exemplary embodiments of a robot control apparatus according to the present invention will be described below in detail with reference to the accompanying drawings. Here, a two-legged humanoid robot is taken as an example of a robot, but the present invention may be applied to other robots such as a four-legged robot.

  FIG. 1 is a diagram illustrating robot walking according to the present embodiment. Here, the right leg of the robot is represented by a solid line, and the left leg of the robot is represented by a dotted line. Here, a plurality of frames Pi defining various postures during the walking motion of the robot are set in advance in the robot control device. To be precise, the frame defines parameters such as the angle of the joints of the robot.

  In FIG. 1, 12 frames from P0 to P11 are set. However, the number of frames is arbitrary. The frame P0 represents the posture before walking where the robot stands alone without falling down. Frames P1, P6, and P11 are postures in which the robot stands alone and does not move without falling down.

  For example, in frame P6, the linear velocity of the robot is 0, the robot stands only with the left leg, and the stride is 0. The robot states of the frame P1 and the frame P11 are the same. That is, in these frames P1 and P11, the linear velocity of the robot is 0, the robot stands only with the right leg, and the stride is 0. A stride of 0 means that both legs of the robot are aligned. In the frames P1, P6, and P11, the robot stands alone and the stride is 0, so these frames are referred to as reference frames.

  Switching from one frame to another is performed by interpolating the postures of all intervening frames. In addition, the postures in the frames other than the reference frame are roughly defined, and these postures are appropriately corrected so that walking is performed stably.

  FIG. 2 is a state transition diagram of each of the frames P0 to P11. The state of the robot transitions from the frame P0 to the reference frame P1 (same as the reference frame P11) or the reference frame P6. Further, the state of the robot can also transition from the reference frame P1 or the reference frame P6 to the frame P0. In the example of FIG. 1, first, the state of the robot transitions from the frame P0 to the reference frame P1. Thereafter, the state of the robot transitions from the reference frame P2 to the reference frame P11 via the reference frame P10.

  As described above, the frame information including the reference frames P1, P6, and P11 where the robot stands alone without falling down is acquired, and the frames P0 to P11 are set so that the robot operation becomes the posture of the reference frames P1, P6, and P11. The state of the robot is controlled by interpolating the posture between the two. As a result, even if the posture of the robot becomes unstable, the posture is stable in the reference frame. That is, the robot can continue the walking motion stably.

  Further, in the reference frames P6 and P11, the robot stops walking and is standing in a stable state. Therefore, the walking step after that, the walking direction is changed, or the movement other than walking is performed. Can be easily started.

  FIG. 3 is a schematic diagram of the robot according to the present embodiment. This robot has two legs: a body 20, a gyro sensor 60 provided on the body, a right leg 30R, and a left leg 30L. Each leg has six joints. That is, the degree of freedom of each leg is 6. The number of joints is arbitrary.

  The joints are pitch hip joint 10, right yaw hip joint 11R, left yaw hip joint 11L, right roll hip joint 12R, left roll hip joint 12L, right pitch hip joint 13R, left pitch hip joint 13L, right pitch knee joint 14R, left pitch knee joint 14L, It includes a right pitch ankle joint 15R, a left pitch ankle joint 15L, a right roll ankle joint 16R, and a left roll ankle joint 16L. A motor (not shown) is incorporated in each joint. The motor of each joint controls the movement of each joint. Each joint has a position sensor (not shown) incorporated therein. The position sensor of each joint detects the movement of each joint, specifically, the rotation angle.

  The pitch hip joint 10 controls the back-and-forth movement (pitching) of the body 20. The right yaw hip joint 11R and the left yaw hip joint 11L cause a movement (yawing) that rotates to the left and right of the robot at the base of each leg.

  The right roll hip joint 12R and the left roll hip joint 12L cause horizontal rotation (rolling) of the robot at the base of each leg. The right pitch hip joint 13R and the left pitch hip joint 13L cause forward and backward rotation (pitching) of the robot at the base of each leg.

  Further, the right pitch knee joint 14R and the left pitch knee joint 14L cause the robot to move back and forth (pitching) in the respective knee portions. The right pitch ankle joint 15R and the left pitch ankle joint 15L cause the robot to move back and forth (pitching) at each ankle portion. The right roll ankle joint 16R and the left roll ankle joint 16L cause a lateral movement (rolling) of the robot in the respective ankle portions.

  In addition, a sole is attached to each leg. FIG. 3 shows a sole 40 attached to the left leg 30L. Four force sensors are incorporated in each sole. The number of force sensors is arbitrary. FIG. 3 shows force sensors 50 a to 50 d provided on the sole 40. These force sensors measure the reaction force that the sole 40 receives from the floor surface. The reaction force measured by the force sensor is used for compliance control and feedback control of the robot motion.

  The gyro sensor 60 measures the rotation angle of the body 20 in the lateral (rolling) direction and the front-rear (pitching) direction. The rotation angle measured by the gyro sensor 60 is used for feedback control of the robot movement.

  FIG. 4 is a time chart for explaining the walking motion of the robot. At time t0, the right roll hip joint 12R, the left roll hip joint 12L, the right roll ankle joint 16R, and the left roll ankle joint 16L perform a rolling operation to tilt both legs to the right side of the robot in order to lift the left leg. When this operation is performed, the movements of the right roll hip joint 12R, the left roll hip joint 12L, the right roll ankle joint 16R, and the left roll ankle joint 16L are slightly different by the gear backlash compensation.

  The amplitude of the rolling motion is determined by trial and error. Alternatively, the amplitude of the rolling operation is determined by performing feedback control so that the evaluation function of the rolling angle of the body 20 is minimized.

  At time t1, the right roll hip joint 12R, the left roll hip joint 12L, the right roll ankle joint 16R, and the left roll ankle joint 16L stop rolling. In order to lift the left leg, that is, to perform the lifting operation of the left leg, the left pitch knee joint 14L rotates to contract the left leg, and the left pitch ankle joint 15L rotates to contract the left ankle. .

  At time t2, the left pitch knee joint 14L and the left pitch ankle joint 15L stop the lifting operation. The right pitch hip joint 13R and the right pitch ankle joint 15R rotate to perform a pitching operation so that the body 20 moves forward.

  Between time t2 and time t3, the right pitch hip joint 13R and the right pitch ankle joint 15R stop the pitching motion.

  At time t3, the left pitch knee joint 14L rotates to extend the left leg to land the left leg on the floor, and the left pitch ankle joint 15L rotates to extend the left ankle.

  At time t4, the left leg lands on the floor. In addition, the right roll hip joint 12R, the left roll hip joint 12L, the right roll ankle joint 16R, and the left roll ankle joint 16L perform a rolling operation to tilt both legs to the left side of the robot in order to lift the right leg. Further, the right pitch hip joint 13R and the right pitch ankle joint 15R rotate so as to return to the original state, that is, the state at time t2. Further, in order to swing the right leg forward, the left pitch hip joint 13L and the left pitch ankle joint 15L rotate to perform a pitching operation so that the body 20 moves forward.

  Between time t4 and time t5, the right pitch hip joint 13R and the right pitch ankle joint 15R return to the original state, and the left pitch hip joint 13L and the left pitch ankle joint 15L stop the pitching operation.

  At time t5, the right roll hip joint 12R, the left roll hip joint 12L, the right roll ankle joint 16R, and the left roll ankle joint 16L stop rolling. Further, in order to lift the right leg, that is, to perform the lifting operation of the right leg, the right pitch knee joint 14R rotates to contract the right leg, and the right pitch ankle joint 15R rotates to contract the right ankle. .

  At time t6, the right pitch knee joint 14R and the right pitch ankle joint 15R stop the lifting operation. Further, the left pitch hip joint 13L and the left pitch ankle joint 15L rotate so as to return to the original state, that is, the state at time t4. As a result, the posture of the robot is set to the posture defined by the reference frame P6.

  Between time t6 and time t7, the left pitch hip joint 13L and the left pitch ankle joint 15L return to the original state. As a result, the posture of the robot is set to the posture defined by the reference frame P6. That is, the linear velocity of the robot is 0, the robot stands only with the left leg, and the stride is 0.

  At time t7, in order to swing the right leg forward, the left pitch hip joint 13L and the left pitch ankle joint 15L rotate and perform a pitching operation so that the body 20 moves forward.

  Between time t7 and time t8, the left pitch hip joint 13L and the left pitch ankle joint 15L stop the pitching motion. At time t8, in order to land the right leg on the floor, the right pitch knee joint 14R rotates to extend the right leg, and the right pitch ankle joint 15R rotates to extend the right ankle.

  At time t9, the right leg lands on the floor. In addition, the right roll hip joint 12R, the left roll hip joint 12L, the right roll ankle joint 16R, and the left roll ankle joint 16L perform a rolling operation to tilt both legs to the right side of the robot in order to lift the left leg. Further, in order to swing the left leg forward, the right pitch hip joint 13R and the right pitch ankle joint 15R rotate to perform a pitching operation so that the body 20 moves forward. Further, the left pitch hip joint 13L and the left pitch ankle joint 15L rotate so as to return to the original state, that is, the state at time t7.

  At time t10, the right roll hip joint 12R, the left roll hip joint 12L, the right roll ankle joint 16R, and the left roll ankle joint 16L stop rolling. Further, in order to lift the left leg, that is, to perform the lifting operation of the left leg, the left pitch knee joint 14L rotates to contract the left leg, and the left pitch ankle joint 15L rotates to contract the left ankle. .

  At time t11, the left pitch knee joint 14L and the left pitch ankle joint 15L stop the lifting operation. Further, the right pitch hip joint 13R and the right pitch ankle joint 15R rotate to return to the original state, that is, the state at time t9. As a result, the posture of the robot is set to the posture defined by the reference frame P11.

  After time t11, the right pitch hip joint 13R and the right pitch ankle joint 15R return to the original state. As a result, the posture of the robot is set to the posture defined by the reference frame P11. That is, the linear velocity of the robot is 0, the robot stands only with the right leg, and the stride is 0. By repeating such a movement, the robot's walking movement is realized.

  FIG. 5 is a functional block diagram of the robot control system according to the present embodiment. This robot control system includes an external terminal device 100 and a robot 110.

  The external terminal device 100 is a personal computer operated by an operator who manages the operation of the robot. The external terminal device 100 communicates with the robot 110. This communication includes the exchange of various types of information.

  The external terminal device 100 transmits frame information of the robot 110 and / or command information to the robot 110 set in advance to the robot 110, information on the state (posture, speed, etc.) of the robot 110 from the robot 110, and the like. Receive. Information obtained from the robot 110 is displayed on a display device (not shown).

The robot 110 is, for example, a two-legged humanoid robot. The robot 110 includes a gyro sensor 111, a gyro sensor control unit 112, joints 113 _{1 to} 113 _n , joint control units 114 _{1 to} 114 _n (n is a natural number), force sensors 115 _{1 to} 115 _m (m is a natural number), force Sensor control units 116 _{1 to} 116 _m , position sensors 117 _{1 to} 117 _n , position sensor control units 118 _{1 to} 118 _n , distance sensors 119 _{1 to} 119 ₂ , distance sensor control units 120 _{1 to} 120 ₂ , communication interface 121, memory 122 and a central control unit 123.

  The gyro sensor 111 has the same function as the gyro sensor 60 shown in FIG. The gyro sensor 111 is provided in the body 20 of the robot 110 and measures the rotation angle of the body 20 in the lateral (rolling) direction and the front-rear (pitching) direction. The gyro sensor control unit 112 controls the function of the gyro sensor 111 and transmits information on the rotation angle measured by the gyro sensor 111 to the central control unit 123.

The joints 113 _{1 to} 113 _n are for moving various joints of the robot 110. A motor (not shown) drives these joints. The joints include the pitch hip joint 10 described in FIG. 3, the right yaw hip joint 11R, the left yaw hip joint 11L, the right roll hip joint 12R, the left roll hip joint 12L, the right pitch hip joint 13R, the left pitch hip joint 13L, the right pitch knee joint 14R, The left pitch knee joint 14L, the right pitch ankle joint 15R, the left pitch ankle joint 15L, the right roll ankle joint 16R, and the left roll ankle joint 16L are included.

The joint control units 114 _{1 to} 114 _n control the operations of the joints 113 _{1 to} 113 _n . In particular, the joint control units 114 _{1 to} 114 _n control the joints 113 _{1 to} 113 _n to rotate at a predetermined angular velocity at a predetermined angle at a predetermined time. The angle, angular velocity, and time are designated by the central control unit 123.

The force sensors 115 _{1 to} 115 _m are provided on the soles of the right leg and the left leg of the robot 110. These force sensors 115 _{1 to} 115 _m measure the reaction force from the floor surface. Further, the force sensors 115 _{1 to} 115 _m have the same functions as the force sensors 50 a to 50 d described with reference to FIG. The force sensor control unit 116 ₁ -116 _m controls the function of the force sensor 115 ₁ to 115 _m, and transmits the information of the reaction force measured by the force sensor 115 ₁ to 115 _m to the central control unit 123.

Position sensors 117 ₁ ~117 _n is attached to each of the joints 113 ₁ to 113 _n, a sensor for detecting the position of the joints 113 ₁ to 113 _n, in particular the rotation angles of the joints 113 ₁ to 113 _n Is detected. Position sensor control unit 118 ₁ - 118 _n controls the operation of the position sensors 117 ₁ ~117 _n, transmits the position information measured by the position sensor 117 ₁ ~117 _n to the central control unit 123.

The distance sensors 119 _{1 to} 119 ₂ are sensors for detecting a step where the walking surface becomes higher, and are provided in front of the right leg and the left leg. The distance sensor control units 120 _{1 to} 120 ₂ control the operation of each of the distance sensors 119 _{1 to} 119 ₂ , and when the walking surface becomes higher by the distance sensors 119 _{1 to} 119 ₂ , the information is sent to the central control unit 123. Send.

  The communication interface 121 communicates with the external terminal device 100. The communication interface 121 performs wireless communication and / or wired communication with the external terminal device 100.

  The memory 122 stores various information. For example, the memory 122 stores information received from the external terminal device 100 and / or information transmitted to the external terminal device 100, and stores information related to the results of various calculations performed by the central control unit 123.

The central control unit 123 controls the robot 110 as a whole. For example, the central control unit 123 calculates the rotation start time, angular velocity, rotation angle, and the like of each joint 113 _{1 to} 113 _n when the robot 110 operates based on the frame information received from the external terminal device 100. The result is transmitted to the joint control units 114 _{1 to} 114 _n .

  Further, the central control unit 123 receives an operation control request for the robot 110 from the external terminal device 100 via the communication interface 121. The motion control request includes a step change request, a walking direction change request, or an execution request for an operation other than walking.

The central control unit 123 executes the request only after the postures of the reference frames P1, P6, and P11 are realized. When executing the request, the central control unit 123 transmits information such as the rotation start time, angular velocity, and rotation angle of the joints 113 ₁ to 113 _n corresponding to the requested motion to the joint control units 114 _{1 to} 114 _n . To do. In the reference frames P1, P6, and P11, the robot 110 stands stably on one foot, so that the above request is executed when the robot 110 stands in a posture corresponding to the reference frames P1, P6, and P11. Is convenient.

Although the central control unit 123 calculates various parameters here, a configuration in which the external terminal device 100 calculates them and controls the robot may be adopted. When such a configuration is adopted, the external terminal device 100 receives information necessary for calculating the rotation start time, angular velocity, rotation angle, and the like from the robot 110, and calculates each parameter based on the received information. . The joint control units 114 _{1 to} 114 _n receive calculation result information from the external terminal device 100 and perform operation control of the robot 110 based on the received information.

  Hereinafter, the robot control process performed by the central control unit 123 will be described in detail. FIG. 6 is a functional block diagram showing the configuration of the central control unit 123. As shown in the figure, the central control unit 123 includes an operation generation unit 123a, a compliance control unit 123b, a feedback control unit 123c, a reflection control unit 123d, and a correction unit 123e.

Based on the frame information received from the external terminal device 100 via the communication interface 121, the motion generation unit 123a performs the rotation start time, angular velocity, rotation angle, etc. of each joint 113 _{1 to} 113 _n when the robot 110 operates. Is a processing unit that calculates and outputs to the correction unit 123e. In addition, the motion generation unit 123a passes phase information indicating which phase the robot 110 is in among the rolling phase, the lifting phase, and the landing (landing) phase to the compliance control unit 123b and the feedback control unit 123c.

The compliance control unit 123b is a processing unit that performs compliance control such as landing motion based on the force sensor data measured by the force sensors 115 _{1 to} 115 _m and the phase information from the motion generation unit 123a, and calculates a compliance control amount. And output to the correction unit 123e.

The feedback control unit 123c is a processing unit that performs gyro feedback control based on the gyro sensor data measured by the gyro sensor 111 and ZMP feedback control based on the force sensor data measured by the force sensors 115 _{1 to} 115 _m . The amount is calculated and output to the correction unit 123e.

The reflection control unit 123d is a processing unit that detects an unexpected event that occurs with respect to a periodic walking motion and performs reflection control, and includes force sensor data and position sensors measured by the force sensors 115 _{1 to} 115 _m . The position sensor data measured by 117 _{1 to} 117 _n , the distance sensor data measured by the distance sensors 119 _{1 to} 119 ₂ , the disturbance to the robot 110 using the gyro sensor data measured by the gyro sensor 111, and the swinging leg Obstacles placed at the landing point, the upper and lower steps of the walking surface, and the fall of the load held by the robot 110 are detected as unexpected events.

  FIG. 7 is an explanatory diagram for explaining the concept of reflection control by the reflection control unit 123d. As shown in the figure, when the reflection control unit 123d detects an unexpected event when the state of the robot 110 changes the state corresponding to each frame, the reflection control unit 123d makes a transition to an unexpected state j. Control is performed to return from the outside state j to the normal state. At this time, the reflection control unit 123d makes a transition to an unexpected state j that minimizes the inertia change of the robot 110. Thus, stable reflection control can be performed by making a transition to an unexpected state j that minimizes the inertia change of the robot 110.

The correction unit 123e corrects the rotation start time, angular velocity, rotation angle, and the like calculated by the motion generation unit 123a for each joint 113 _{1 to} 113 _n with the outputs of the compliance control unit 123b, the feedback control unit 123c, and the reflection control unit 123d. And a processing unit that outputs commands to the motors of the joints 113 _{1 to} 113 _n .

  Next, regarding the details of the reflection control by the reflection control unit 123d, a large disturbance to the robot 110, an obstacle placed at the landing point of the swinging leg, the upper and lower steps of the walking surface, and the weight of the luggage held by the robot 110 suddenly The change will be described in order as an unexpected event.

  FIG. 8 is an explanatory diagram for explaining reflection control for a large disturbance. Here, the large disturbance is a large force received by the robot 110 being pressed from the outside. In FIG. 8, a normal ZMP trajectory when the robot 110 is standing on one leg with the right leg while walking is indicated by a thick arrow.

  When the robot 110 receives a disturbance during walking, the ZMP deviates from the normal trajectory depending on the direction of the disturbance. However, the reflection control unit 123d has any of the areas a, b, c, and d shown in FIG. The robot 110 is controlled so that the state of the robot 110 transitions to the state of the frame previously associated with each of the areas a, b, c, and d to which the ZMP has moved. For example, when the robot 110 is pushed from the right rear, the ZMP moves to the area a, and the reflection control unit 123d performs control so that the state of the robot 110 transitions to a frame state previously associated with the area a. . The association between each region and the frame is performed by learning.

Further, the reflection control unit 123d calculates the ZMP using the force sensor data in order to determine whether the ZMP has moved to any of a, b, c, and d shown in FIG. FIG. 9 is a diagram illustrating a ZMP calculation method by the reflection control unit 123d. As shown in the figure, the reflection control unit 123d calculates ZMP (x _m , y _m ) using the _four force sensor data F _{1 to} F ₄ and the coordinates to which the force sensor is attached.

  FIG. 10 is a flowchart showing the procedure of the walking control process for a large disturbance. As shown in the figure, in this walking control process, the central control unit 123 records the ZMP when the robot 110 is standing on one leg (step S101), and the | measurement value | of the gyro sensor is smaller than a predetermined threshold value. Whether or not (step S102). As a result, when the | measured value | of the gyro sensor is smaller than a predetermined threshold, since a large disturbance has not occurred, normal walking control is performed (step S103).

  On the other hand, if the | measured value | of the gyro sensor is not smaller than the predetermined threshold value, a large disturbance has occurred, so the reflection control unit 123d stops the walking motion of the robot 110 (step S104) and moves the ZMP. It is specified which region is a to d shown in FIG. 8 (step S105).

  Then, the reflection control unit 123d specifies a frame corresponding to the specified region (step S106), generates an operation so as to shift to the specified frame from the current position measured by the position sensor, and outputs the operation to the correction unit 123e. (Step S107). Then, the central control unit 123 activates compliance control and feedback control (step S108).

  As described above, the reflection control unit 123d generates a motion of the robot 110 that stops the walking motion of the robot 110 and shifts to the frame specified based on the movement position of the ZMP when a large disturbance occurs. Even when a large disturbance occurs, the robot 110 can be made to respond.

  FIG. 11 is a diagram illustrating an example of reflection control for a large disturbance. FIG. 2A shows an example in which the robot 110 is pushed diagonally forward while standing on the right leg and the ZMP moves to the region a. FIG. 2B shows an example in which the robot 110 is pushed backward when the right leg is standing and the ZMP moves to the region b. FIG. 3 (3) shows an example where the robot 110 is pushed diagonally backward while standing on the right leg and the ZMP moves between the regions b and d. In FIG. 3C, the reflection control unit 123d generates a frame that interpolates a frame corresponding to the region b and a frame corresponding to the region d, and controls the robot 110 to shift to the generated frame.

  FIG. 12 and FIG. 13 are explanatory diagrams for explaining the reflection control for the obstacle placed at the landing point of the swinging leg. As shown in FIG. 12, when the obstacle is detected by the swing leg force sensor, the reflection control unit 123d controls the swing leg to return to the original position. Therefore, the ZMP returns from the swing leg to the support leg as shown in FIG. In order to detect an obstacle with the force sensor of the swing leg, the gain for compliance control is set large.

  FIG. 14 is a flowchart showing a processing procedure of walking control processing for an obstacle placed at a landing point. As shown in the figure, in this walking control process, the central control unit 123 sets a large gain for compliance control (step S201), and normal walking control is performed while no obstacle is detected (No in step S202). Is performed (step S203).

  On the other hand, when the central control unit 123 detects an obstacle (Yes in step S202), the reflection control unit 123d instructs the motion generation unit 123a to return the waist to the support leg and smoothly move other joints. Stop (step S204).

  Then, the reflection control unit 123d instructs the motion generation unit 123a to increase the gate time (walking motion time), and return the swing leg (step S205). Thereafter, the central control unit 123 continues the operation according to the host control (step S206). Note that during the operations in steps S204 and S205, the compliance control and the feedback control are in an activated state.

  As described above, when there is an obstacle at the landing point of the swinging leg, the reflection control unit 123d instructs the motion generation unit 123a to return the waist to the support leg and smoothly stop the movement of other joints. In addition, by increasing the gate time and returning the swing leg, the robot 110 can be made to respond to an obstacle.

  FIG. 15 is a diagram illustrating an example of reflection control for an obstacle. As shown in the figure, the swinging left leg is in contact with an obstacle when landing and is returned to its original position.

  FIG. 16 is an explanatory diagram for explaining the reflection control when the walking surface is lowered. As shown in the figure, when the walking surface is lowered, an additional operation is required to cope with the decrease in height. Specifically, it is necessary to extend the gate time and continue the landing operation to the limit of hardware.

  FIG. 17 is a flowchart showing a processing procedure of walking control processing when the walking surface is lowered. As shown in the figure, in this walking control process, the central control unit 123 sets a large gain for compliance control (step S301) and normal walking control while landing at the scheduled time (step S302, positive). Is performed (step S303).

  On the other hand, when the central control unit 123 detects that the landing has not been performed at the scheduled time (No at Step S302), the reflection control unit 123d instructs the operation generation unit 123a to continue the landing operation and extend the gate time. (Step S304), it is determined whether or not the sole of the foot touches the floor within the hardware limit (Step S305).

  As a result, if the sole of the foot touches the floor within the hardware limit, the process proceeds to step 303 where the central control unit 123 performs normal walking control, and the sole of the foot does not touch the floor within the hardware limit. In this case, the reflection control unit 123d instructs the action generation unit 123a to return the swing leg (step S306). Thereafter, the central control unit 123 continues the operation according to the upper control (step S307). Note that the compliance control and the feedback control are in an activated state during the operations in steps S304 to S306.

  As described above, when the walking surface becomes low, the reflection control unit 123d instructs the motion generation unit 123a to continue the landing motion and extend the gate time, thereby causing the robot 110 to respond to the decrease in the walking surface. it can.

  FIG. 18 is a diagram illustrating joint angles when the walking surface is lowered. In the figure, a indicates the angle of the left pitch knee joint 14L, b indicates the angle of the left pitch ankle joint 15L, and c indicates the angle of the left pitch hip joint 13L. Further, d indicates an additional operation of the supporting leg joint.

  FIG. 19 is a diagram illustrating an example of reflection control when the walking surface is lowered. As shown in the figure, even when the walking surface is lowered, the robot 110 can stably walk by continuing the landing motion.

  FIG. 20 is an explanatory diagram for explaining the reflection control when the walking surface is high. As shown in the figure, when the walking surface becomes high, it is necessary to increase the lifting of the legs to cope with the increase in height.

  FIG. 21 is a flowchart showing a processing procedure of walking control processing when the walking surface becomes high. As shown in the figure, in this walking control process, the central control unit 123 sets a large gain for compliance control (step S401), and while the distance sensor does not detect ascending of the walking surface (No at step S402). Then, normal walking control is performed (step S403).

  On the other hand, when the distance sensor detects the ascending of the walking surface (Yes at Step S402), the reflection control unit 123d instructs the motion generation unit 123a to increase the leg lift and extend the gate time (Step S404). . Then, when the floor touches the sole of the foot (Yes at Step S405), the motion generation unit 123a is instructed to increase the stride (Step S406). Then, the central control unit 123 returns to step S403 to perform normal walking control.

  The reflection control unit 123d increases the lifting of the leg within the hardware limit while the back of the foot does not touch the floor, and the sole does not touch the floor even if the leg is lifted to the hardware limit. Instructs the motion generator 123a to return the swing leg.

  As described above, when the walking surface becomes high, the reflection control unit 123d instructs the motion generation unit 123a to increase the lifting of the leg and extend the gate time. 110 can be made to correspond.

  FIG. 22 is an explanatory diagram for explaining the reflection control when the robot 110 drops a load. As shown in FIG. 22, when the load carried by the robot 110 is dropped, a moment of backward rotation is generated in the robot 110 as shown in FIG. Therefore, in the reflection control, a motion that cancels the moment of back rotation is generated on the waist and ankle.

  FIG. 23 is a flowchart illustrating a processing procedure of the walking control process when the robot 110 drops a load. As shown in the figure, in this walking control process, the central control unit 123 determines whether or not | measured value | of the gyro sensor is smaller than a predetermined threshold value (step S501). As a result, when the | measured value | of the gyro sensor is smaller than the predetermined threshold value, the robot 110 does not drop the load, and thus performs normal walking control (step S502).

On the other hand, if the | measured value | of the gyro sensor is not smaller than a predetermined threshold value, the robot 110 has dropped the load, so the reflection control unit 123d reduces the gain of the gyro feedback control and displays it on the ankle and waist. The motion generation unit 123a is instructed to generate the motion shown in FIG. 24 (step S503). In FIG. 24, θ _w indicates the movement of the pitch hip joint 10 generated by the reflection control, and θ _a indicates the movement of the pitch ankle joint by the reflection control.

  Then, the reflection control unit 123d restores the gain of the gyro feedback control (Step S504), and instructs the motion generation unit 123a to perform normal walking control or stop (Step S505).

  As described above, when the robot 110 drops the load, the reflection control unit 123d instructs the motion generation unit 123a to reduce the gain of the ZMP feedback control and generate the movement shown in FIG. 24 at the ankle and the waist. Thus, the robot 110 can be put under control.

FIG. 25 is a diagram illustrating an example of reflection control when the robot 110 drops a load. As shown in the figure, by the reflection control, the robot 110 can maintain a stable state even when a load is dropped. FIG. 26 is a diagram illustrating joint angles when the robot 110 drops a load. In the figure, V _ZMP indicates the ZMP feedback of the pitch hip joint 10 and U _ZMP indicates the ZMP feedback of the pitch ankle joint.

  As described above, in this embodiment, when the central control unit 123 detects an unexpected event such as a large disturbance, an obstacle, a vertical step on the walking surface, or a fall of the transported bag during the walking control of the robot 110. The reflection control unit 123d issues an operation instruction to the operation generation unit 123a so that the change in inertia of the robot 110 is minimized in response to the event, and the operation generation unit 123a generates a correction operation, or the reflection control unit 123d Since the correction operation is directly generated and output to the correction unit 123e, the robot 110 can be made to respond to a change in the walking environment.

  Although the robot control apparatus has been described in the present embodiment, a robot control program having the same function can be obtained by realizing the configuration of the robot control apparatus with software. A computer that executes this robot control program will be described.

  FIG. 27 is a functional block diagram illustrating a configuration of a computer that executes the robot control program according to the present embodiment. As shown in the figure, the computer 200 includes a RAM 210, a CPU 220, a flash memory 230, a USB interface 240, and a COM interface 250.

  The RAM 210 is a memory that stores a program, a program execution result, and the like, and corresponds to the memory 122 shown in FIG. The CPU 220 is a central processing unit that reads a program from the RAM 210 and executes the program. The flash memory 230 is a non-volatile memory that stores programs and data, and the USB interface 240 is an interface for connecting the computer 200 to joints and sensors. The COM interface 250 is an interface for communicating with the external terminal device 100, and corresponds to the communication interface 121 shown in FIG. The robot control program 231 executed in the computer 200 is read from the flash memory 230 and executed by the CPU 220.

(Supplementary note 1) A robot control device that controls walking of a robot,
A walking motion control means for generating control information based on postures at a plurality of different times including at least a reference posture where the robot stands alone without falling, and controlling the robot to perform a predetermined walking motion; ,
A reflection control means for detecting a change in the walking environment that makes the continuation of the predetermined walking motion impossible, and correcting the control of the walking motion by the walking motion control means in response to the detected change;
A robot control device comprising:

(Appendix 2) The reflection control means, when detecting the presence of an obstacle placed at the place where the robot leg lands, performs a rolling operation to the support leg and an operation to return the swinging leg to the position before the swing. The robot control apparatus according to appendix 1, characterized in that the walking motion control means is instructed.

(Appendix 3) When the reflection control means detects that the walking surface on which the robot walks is lowered, the reflection control means instructs the walking action control means to extend the walking action time and continue the landing action. The robot control apparatus according to appendix 1.

(Supplementary Note 4) When the reflection control means detects that the walking surface on which the robot walks becomes high, the reflection control means instructs the walking movement control means to lengthen the walking movement time and raise the landing leg. The robot control apparatus according to Supplementary Note 1.

(Supplementary Note 5) When the weight of the load held by the robot suddenly changes, the reflection control means reduces the gain of the gyro feedback control to a predetermined value and controls the walking motion so as to rotate the ankle and the waist. The robot control apparatus according to appendix 1, wherein the robot control apparatus instructs the means.

(Supplementary note 6) The robot control according to supplementary note 5, wherein the reflection control means detects a fall of a load held by the robot based on a measurement value of a gyro sensor attached to a body portion of the robot. apparatus.

(Appendix 7) The reflection control means corrects the control so as to stop the walking motion and shift to a predetermined posture with respect to the direction of the pressing force when detecting the pressing force of the robot. The robot control apparatus according to appendix 1.

(Additional remark 8) The said reflection control means detects the change of a walking environment by enlarging the gain of a compliance control compared with the case where it controls so that the said predetermined walking action may be performed. 3. The robot control device according to 3 or 4.

(Supplementary note 9) A robot control method for controlling walking of a robot,
A walking motion control step for generating control information based on postures at a plurality of different times including at least a reference posture where the robot stands alone without falling down, and controlling the robot to perform a predetermined walking motion; ,
A reflection control step of detecting a change in a walking environment that makes it impossible to continue the predetermined walking motion, and correcting the control of the walking motion by the walking motion control step in response to the detected change;
The robot control method characterized by including.

(Supplementary Note 10) A robot control program for controlling walking of a robot,
A walking motion control procedure for generating control information based on postures at a plurality of different time points including at least a reference posture where the robot stands alone without falling, and controlling the robot to perform a predetermined walking motion; ,
A reflection control procedure that detects a change in the walking environment that makes the predetermined walking motion impossible to continue, and corrects the control of the walking motion by the walking motion control procedure in response to the detected change;
A robot control program for causing a computer to execute.

  As described above, the robot control apparatus according to the present invention is useful for robot walking control, and is particularly suitable for a robot walking in a changing environment.

It is a figure explaining the robot walking which concerns on a present Example. It is a state transition diagram of each frame P0-P11. It is the schematic of the robot which concerns on a present Example. It is a time chart explaining the walking movement of a robot. It is a functional block diagram of the robot control system according to the present embodiment. It is a functional block diagram which shows the structure of a central control part. It is explanatory drawing for demonstrating the view of reflection control by a reflection control part. It is explanatory drawing for demonstrating the reflection control with respect to a big disturbance. It is a figure which shows the calculation method of ZMP by a reflection control part. It is a flowchart which shows the process sequence of the walk control process with respect to a big disturbance. It is a figure which shows the example of reflection control with respect to a big disturbance. It is a figure which shows the obstruction placed in the landing point of the leg which swung. It is a figure which shows the return of ZMP to a support leg. It is a flowchart which shows the process sequence of the walk control process with respect to the obstruction placed at the landing point. It is a figure which shows the example of reflection control with respect to an obstruction. It is explanatory drawing for demonstrating reflection control in case a walking surface becomes low. It is a flowchart which shows the process sequence of the walk control process in case a walking surface becomes low. It is a figure which shows the angle of a joint in case a walking surface becomes low. It is a figure which shows the example of reflection control in case a walking surface becomes low. It is explanatory drawing for demonstrating reflection control in case a walking surface becomes high. It is a flowchart which shows the process sequence of the walk control process in case a walk surface becomes high. It is explanatory drawing for demonstrating reflection control when a robot drops a load. It is a flowchart which shows the process sequence of the walk control process when a robot drops a load. It is a figure which shows the motion produced | generated by the ankle and the waist. It is a figure which shows the example of reflection control when a robot drops a load. It is a figure which shows the angle of the joint when a robot drops a load. It is a functional block diagram which shows the structure of the computer which performs the robot control program which concerns on a present Example.

Explanation of symbols

10 pitch hip joint 11R right yaw hip joint 11L left yaw hip joint 12R right roll hip joint 12L left roll hip joint 13R right pitch hip joint 13L left pitch hip joint 14R right pitch knee joint 14L left pitch knee joint 15R right pitch ankle joint 15L left pitch ankle joint 16R right roll ankle joint 16L left roll ankle joint 20 body 30R right leg 30L left leg 40 feet back 50a~50d, 115 ₁ ~115 _m force sensor 60,111 gyro sensor 100 external terminal device 110 robot 112 gyro sensor control unit 113 ₁ to 113 _n joints 114 _{1 to} 114 _n joint control units 116 _{1 to} 116 _m force sensor control units 117 _{1 to} 117 _n position sensors 118 _{1 to} 118 _n position sensor control units 119 _{1 to} 119 ₂ distance sensors 120 _{1 to} 120 ₂ distance sensor control Part 121 communication interface Face 122 memory 123 central control unit 123a operates generating unit 123b compliance control unit 123c feedback controller 123d reflection control unit 123e correction section 200 computer 210 RAM
220 CPU
230 Flash memory 231 Robot control program 240 USB interface 250 COM interface

Claims (5)

A robot control device for controlling walking of a robot,
A walking motion control means for generating control information based on postures at a plurality of different times including at least a reference posture where the robot stands alone without falling, and controlling the robot to perform a predetermined walking motion; ,
When a value calculated from a point where a combined vector of inertial force and gravity acting on the robot intersects with the sole of the foot deviates from the predetermined value, the value is associated in advance with the calculated value. A reflection control means for correcting the control of the walking motion by the walking motion control means so as to return to the posture in which the predetermined walking motion can be continued after the transition to the posture that minimizes the inertia change ;
A robot control device comprising:   The reflection control means, when detecting the presence of an obstacle placed at a place where the robot's leg lands, performs the walking operation so as to perform a rolling operation to the support leg and an operation to return the swinging leg to the position before the swing. 2. The robot control apparatus according to claim 1, wherein the controller is instructed.   2. The reflection control means, when detecting that the walking surface on which the robot walks is lowered, instructs the walking movement control means to extend the walking movement time and continue the landing movement. The robot control device described in 1.   The reflection control means, when detecting that the walking surface on which the robot walks becomes high, instructs the walking movement control means to lengthen the walking movement time and raise the landing leg. The robot control apparatus according to 1.   The reflection control means instructs the walking movement control means to reduce the gain of the gyro feedback control to a predetermined value and rotate the ankle and the waist when the weight of the load held by the robot suddenly changes. The robot control apparatus according to claim 1.

Images