JP2011224757A - Device for controlling two-leg mobile robot and gait generating device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent a large moment from acting on a joint of a base end part of a leg body carrying out floor-leaving and floor-landing actions when making a two-leg mobile robot carry out one-leg hopping action.SOLUTION: When making the two-leg mobile robot 1 carry out the one-leg hopping action, the target movement (a target posture of an upper body 24) of the robot 1 is generated in such a way that a base end part of a leg body 2 (the leg body 2 at a free leg side) different from a leg body 2 at a support leg side is present at a relatively higher position than the base end part of the leg body 2 at the support leg side, in a floor landing state after floor-leaving of the leg body 2 (the leg body 2 at the support leg side) carrying out the floor-leaving and floor-landing actions.

Description

  The present invention relates to a control device and a gait generator for a biped mobile robot.

  In a biped mobile robot having two legs connected to the upper body via a joint, such as a humanoid robot, a series of actions (hereinafter, simply referred to as a floor movement operation of each leg and a subsequent landing operation) A technique for moving a robot by performing switching between the two legs alternately (sometimes referred to as leaving or landing) (ie, by walking or running the robot) has been conventionally applied. It has been proposed by the applicant or the like (see, for example, Patent Document 1).

Japanese Patent No. 3675788

  By the way, the form of the movement of the two legs required in the biped mobile robot is not limited to the movement performed by alternately switching the above-mentioned getting-off / landing operation between the two legs. For example, when a lateral external force is applied to the robot due to contact between the biped mobile robot and another object or the like, one of the two legs is avoided in order to avoid the posture of the robot from collapsing. It is preferable to cause the robot to perform an operation (hereinafter also referred to as a one-leg hopping operation) of performing the above-mentioned floor leaving / landing operation only once or a plurality of times only with the other leg while the leg is left. There is a case. It is also conceivable to cause the robot to perform a one-leg hopping operation according to the request of the robot operator.

  On the other hand, biped mobile robots are generally bilaterally symmetric robots. Therefore, when the above-mentioned leaving and landing operations are alternately switched between two legs (when the robot is walking or running). The posture of the upper body (base) of the robot when the robot is viewed in a lateral plane (frontal plane) is usually the posture when the robot is standing upright or a posture close to the posture (viewed in the lateral plane). The posture in which the direction of the trunk axis of the upper body is substantially vertical).

  However, when causing the robot to perform the one-leg hopping operation, if the posture of the upper body of the robot is maintained in the same posture as the walking operation or the running operation, the following inconvenience occurs. It became clear by examination.

  That is, when the biped mobile robot performs the above-described one-leg hopping operation, in the state where one leg that performs the floor leaving / landing operation has landed after leaving the floor, the overall center of gravity of the robot, the one leg It is desirable to be located almost directly above the ground contact surface.

  In this case, if the trunk axis of the upper body of the robot viewed in the lateral plane is oriented in a substantially vertical direction (when the upper body is in the upright position or a position close to the above), the user must land The position of the base end portion (the connecting portion with the upper body) of the one leg that is present is located at a position that is laterally shifted from immediately above the ground contact surface of the one leg. This is because, in a biped mobile robot with a bilaterally symmetric structure, the center of gravity of the robot in a state where the trunk axis of the upper body of the robot is substantially in the vertical direction is the lateral direction (left-right direction) of the robot, This is because it exists near the center between the base ends of the two legs.

  Therefore, in particular, during the period immediately after the one leg leaves the floor after landing, it is caused by the gravity acting on the robot and the inertial force in the vertical direction (in other words, the gravity and inertial force Due to the floor reaction force that balances the resultant force), a relatively large moment acts around the axis in the front-rear direction of the robot (around the roll axis) on the joint at the base end of the one leg.

  As a result, there is an inconvenience that the joint of the base end portion of the one leg or the actuator for driving the same is increased and the weight is increased.

  The present invention has been made in view of such a background, and when a bipedal mobile robot performs a one-leg hopping operation, a large moment acts on the joint of the base end portion of the leg that performs the floor leaving and landing operations. An object of the present invention is to provide a control device for a bipedal mobile robot and a gait generator that can prevent this from occurring.

  The control device for a biped mobile robot according to the present invention is a biped mobile robot having two legs whose base ends are connected to the upper body via joints. A biped mobile robot including a one-leg hopping operation in which one of the two legs is removed from the floor and the other leg is used to perform the leaving action and the subsequent landing action. In the control device, when the robot performs the one leg hopping operation, the other leg (hereinafter referred to as a supporting leg side leg) may be in the landing state after the other leg leaves the floor. ) At least the target motion of the robot so that the base end of the one leg (hereinafter sometimes referred to as the free leg side leg) is located at a relatively higher position than the base end of A target gait generating means for generating a target gait including Characterized in that it comprises a joint control means for controlling an operation of each joint of the robot in accordance with the desired gait the generated (first invention).

  According to the first aspect of the invention, the target gait generating means performs landing after leaving the supporting leg side leg that performs the leaving action and the landing action when the robot performs the one leg hopping action. In the state, the target gait including at least the target motion of the robot so that the base end of the free leg side leg exists at a position relatively higher than the base end of the support leg side leg. Is generated.

  Here, the posture of the upper body of the robot in a state where there is no difference between the height of the base end portion of the supporting leg side leg and the height of the base end portion of the free leg side leg, the robot is in an upright posture. Corresponds to the posture of the upper body while standing.

  For this reason, the posture of the upper body in the target gait generated by the target gait generating means in the landing state after leaving the supporting leg-side leg is higher than the posture of the upper body in the upright posture. The portion closer to the upper part of the body is inclined so as to approach directly above the base end of the supporting leg side leg.

  In the posture of the body tilted in this way, the horizontal distance between the center of gravity of the entire robot and the base end of the supporting leg side leg is the posture in which the body posture corresponds to the upright posture. This is shorter than the case of.

  In the first invention, motion control of each joint of the robot is performed according to the target gait generated as described above.

  As a result, in the landing state of the supporting leg side leg after leaving the floor in the one-leg hopping operation of the robot, it is caused by the resultant force of gravity acting on the robot and the inertial force in the vertical direction, or the floor reaction force that balances the resultant force. Thus, it is possible to suppress a large moment from acting on the joint at the base end portion of the supporting leg side leg.

  Therefore, according to the first aspect of the present invention, it is possible to prevent a large moment from acting on the joint of the base end portion of the leg that performs the floor leaving / landing operation when the biped mobile robot performs the one leg hopping operation. it can.

  Further, the control device for a biped mobile robot of the present invention is a biped mobile robot having two legs whose base ends are connected to the upper body via joints. Two-legged movement including a one-leg hopping operation in which one of the two legs is removed from the floor and only the other leg is used to perform a leaving action and a subsequent landing action. When the robot performs the one-leg hopping operation, the robot's overall center-of-gravity point and the base of the other leg in the landing state after the other leg leaves the floor. Target gait generation that generates a target gait that includes at least the target motion of the robot so that a horizontal distance to the end is shorter than the horizontal distance in a state where the robot stands in an upright posture Means and said raw Depending on the desired gait, which is characterized in that it comprises a joint control means for controlling an operation of each joint of the robot (second invention).

  According to the second aspect of the present invention, when the desired gait generating means causes the one-leg hopping operation to be performed by the robot, in the landing state after the other leg (supporting leg-side leg) leaves the floor, The horizontal distance between the center of gravity of the entire robot and the base end of the other leg is shorter than the horizontal distance when the robot stands upright. A target gait including at least the target motion is generated.

  For this reason, in the target gait generated in the landing state after the support leg-side leg leaves the floor, as in the first invention, the center of gravity of the entire robot and the base end of the support leg-side leg The horizontal distance between the upper body and the body is shorter than when the body posture is a posture corresponding to the upright posture.

  In the second invention, motion control of each joint of the robot is performed according to the target gait thus generated.

  As a result, in the landing state after leaving the support leg side leg in the one leg hopping operation, due to the resultant force of gravity acting on the robot and the inertia force in the vertical direction, or the floor reaction force that balances the resultant force, It is suppressed that a big moment acts on the joint of the base end part of the supporting leg side leg.

  Therefore, according to the second aspect of the present invention, it is possible to prevent a large moment from acting on the joint of the base end portion of the leg that performs the floor leaving / landing operation when the biped mobile robot performs the one leg hopping operation. it can.

  In the first invention or the second invention, the target motion generated by the target gait generating means in the landing state after the other leg leaves the floor is the center of gravity of the entire robot and the other leg. It is preferable that the base end portion of the body is a target motion that is arranged so as to be aligned vertically above the ground contact surface of the other leg (third invention).

  According to the third aspect of the invention, the moment acting on the base end portion can be made substantially zero, and the load on an actuator (for example, a motor) that drives the joint of the base end portion can be reduced.

  In addition, the gait generator for a biped mobile robot of the present invention is a biped mobile robot having two legs whose base ends are connected to the upper body via joints, and the robot can perform an operation. 2 including a one-leg hopping operation in which a floor leaving operation and a subsequent landing operation are performed only by the other leg in a state where one of the two legs is removed from the floor. A gait generating device for a legged mobile robot, wherein when generating a target gait including at least a target motion of the robot for causing the robot to perform the one-leg hopping motion, Target motion generating means for generating the target motion of the robot so that the base end of the one leg exists at a relatively higher position than the base end of the other leg in the landing state (No. 1) Invention).

  According to the fourth aspect of the invention, when the target motion generating means generates a target gait including at least the target motion of the robot for causing the robot to perform the one leg hopping motion, the other leg ( In the landing state after leaving the support leg side leg), the base end portion of the one leg (leg leg side leg) is relatively higher than the base end portion of the support leg side leg. The target motion of the robot is generated so as to exist at the position.

  For this reason, in the target motion generated in the landing state after the support leg-side leg leaves the floor, as in the first invention, the overall center of gravity of the robot and the base end of the support leg-side leg The horizontal distance between the two is shorter than when the posture of the upper body is a posture corresponding to the upright posture.

  As a result, in the landing state after leaving the support leg side leg in the one leg hopping operation, due to the resultant force of gravity acting on the robot and the inertia force in the vertical direction, or the floor reaction force that balances the resultant force, A target motion that can suppress a large moment from acting on the joint at the base end of the supporting leg side leg is generated.

  Therefore, according to the fourth aspect of the present invention, it is possible to prevent a large moment from acting on the joint of the base end portion of the leg that performs the floor leaving / landing operation when the biped mobile robot performs the one leg hopping operation. A possible target motion can be generated.

In addition, the gait generator for a biped mobile robot of the present invention is a biped mobile robot having two legs whose base ends are connected to the upper body via joints, and the robot can perform an operation. 2 including a one-leg hopping operation in which a floor leaving operation and a subsequent landing operation are performed only by the other leg in a state where one of the two legs is removed from the floor. A gait generating device for a legged mobile robot, wherein when generating a target gait including at least a target motion of the robot for causing the robot to perform the one-leg hopping motion, In the landing state, the horizontal distance between the center of gravity of the entire robot and the base end of the other leg is shorter than the horizontal distance when the robot stands upright. So that the robot Characterized in that it comprises a desired motion generating means for generating a target motion (fifth invention)
According to the fifth aspect of the present invention, when the target motion generating means generates a target gait including at least the target motion of the robot for causing the robot to perform the one leg hopping motion, the other leg ( In the landing state after leaving the support leg-side leg), the horizontal distance between the center of gravity of the entire robot and the base end of the other leg is such that the robot stands upright. The target motion of the robot is generated so as to be shorter than the horizontal distance at.

  For this reason, in the target motion generated in the landing state after the support leg-side leg leaves the floor, as in the first invention, the overall center of gravity of the robot and the base end of the support leg-side leg The horizontal distance between the two is shorter than when the posture of the upper body is a posture corresponding to the upright posture.

  As a result, in the landing state after leaving the support leg side leg in the one leg hopping operation, due to the resultant force of gravity acting on the robot and the inertia force in the vertical direction, or the floor reaction force that balances the resultant force, A target motion that can suppress a large moment from acting on the joint at the base end of the supporting leg side leg is generated.

  Therefore, according to the fifth aspect of the present invention, it is possible to prevent a large moment from acting on the joint of the base end portion of the leg that performs the floor leaving / landing operation when the biped mobile robot performs the one leg hopping operation. A possible target motion can be generated.

  In the fourth invention or the fifth invention, the target motion generated by the target gait generating means when the robot performs the one-leg hopping motion is the center of gravity of the entire robot and the other leg. It is preferable that the base end of the body is a target motion that is arranged so as to be lined up and down immediately above the ground contact surface of the other leg (the sixth invention).

  According to the sixth aspect of the invention, as in the third aspect of the invention, the moment acting on the base end can be made substantially zero, and the load of an actuator (for example, a motor) that drives the joint of the base end can be reduced. Can be reduced.

The figure which shows the outline of the whole structure of the bipedal mobile robot in embodiment of this invention. The block diagram which shows the structure of the control unit with which the robot of FIG. 1 was equipped. The block diagram which shows the functional structure of the control unit of FIG. Explanatory drawing which illustrates a one leg hopping gait. The figure which shows the dynamic model used for gait generation. The flowchart which shows the main routine process of the gait generator shown in FIG. The figure which illustrates the landing position and posture of a foot in a one-leg hopping gait, and a support leg coordinate system. 7 is a flowchart showing a subroutine process of S022 shown in FIG. The graph which shows the example of a setting of the reference body posture angle of the X-axis periphery (roll direction) in a one leg hopping gait. The figure for demonstrating the setting guideline of the reference | standard body posture angle shown in FIG. The figure which shows the operation state of the robot corresponding to the reference | standard body posture angle shown in FIG. The graph which shows the example of a setting of the desired floor reaction force vertical component in a normal gait. The graph which shows the example of a setting of the floor reaction force horizontal component tolerance | permissible_range in a normal gait. 7 is a flowchart showing a subroutine process of S024 shown in FIG. 15 is a flowchart showing a subroutine process of S208 shown in FIG. 16 is a flowchart showing a subroutine process of S306 shown in FIG. 17 is a flowchart showing a subroutine process of S412 shown in FIG. The graph which shows the example of a setting of the pattern of the body posture restoring moment ZMP conversion value set by S518 of FIG. 7 is a flowchart showing a subroutine process of S026 shown in FIG. The graph which shows the setting example of the floor reaction force vertical component of a gait this time. The graph which shows the setting example of the floor reaction force horizontal component allowable range of a gait this time. 7 is a flowchart showing subroutine processing of S028 shown in FIG. 23 is a flowchart showing a subroutine process of S702 shown in FIG. The flowchart which shows the subroutine processing of S806 shown in FIG. The flowchart which shows the subroutine processing of S912 shown in FIG. The graph which shows the example of a setting of the ZMP correction amount used by S716 of FIG. The graph which shows the example of a setting of the reference | standard body posture angle | corner of the X-axis periphery (roll direction) in the case of changing a walking gait or a running gait from a one leg hopping gait. The graph which shows the example of a setting of the reference | standard body posture angle of the X-axis periphery (roll direction) in the case of changing to a one-leg hopping gait from a walking gait or a running gait.

  Hereinafter, an embodiment of the present invention will be described.

  As shown in FIG. 1, a bipedal mobile robot 1 (hereinafter simply referred to as a robot 1) of the present embodiment includes an upper body 24 as a base and a pair of left and right legs (leg portions) extending from the upper body 24. Link) 2R, 2L is a robot with a symmetrical structure. In the description of the present embodiment, the symbols “R” and “L” correspond to the right leg and the left leg, respectively.

  The upper body 24 is connected to the base ends (upper end portions) of both legs 2R, 2L via a hip joint (hip joint), which will be described later, and the ground surface of the legs 2R, 2L touches the floor surface. Supported upward.

  Both legs 2R and 2L have the same structure, and each has six joints. The six joints are, in order from the upper body 24 side, joints 10R and 10L for rotating the hip (crotch) (for rotating in the yaw direction with respect to the upper body 24), and the roll direction of the hip (crotch) (around the X axis). Joints 12R, 12L for rotation, joints 14R, 14L for rotation in the hip (crotch) pitch direction (around the Y axis), joints 16R, 16L for rotation in the pitch direction of the knee, It consists of joints 18R, 18L for rotation in the pitch direction and joints 20R, 20L for rotation in the roll direction of the ankle part.

  The X axis, the Y axis, and the Z axis mean three coordinate axes of a support leg coordinate system described later. In this support leg coordinate system, the X-axis direction and the Y-axis direction are two axis directions orthogonal to each other on the horizontal plane, the X-axis direction is the front-rear direction (roll axis direction) of the robot 1, and the Y-axis direction is the left-right direction of the robot This corresponds to the direction (pitch axis direction). The Z-axis direction is the vertical direction (gravity direction) and corresponds to the vertical direction (yaw axis direction) of the robot 1.

  The joints 10R (L), 12R (L), and 14R (L) of each leg 2R (L) constitute a three-degree-of-freedom hip joint (hip joint), and the joint 16R (L) constitutes a one-degree-of-freedom knee joint. An ankle joint having two degrees of freedom is configured by the joints 18R (L) and 20R (L).

  The hip joints (hip joints) 10R (L), 12R (L), 14R (L) and the knee joint 16R (L) are connected by a thigh link 28R (L), and the knee joint 16R (L) and the ankle joint 18R. (L) and 20R (L) are connected by a crus link 30R (L). Also, a foot 22R (L) constituting the tip (lower end) of each leg 2R (L) is attached to the lower part of the ankle joints 18R (L), 20R (L) of each leg 2R (L). It is worn. Further, the upper end (base end) of each leg 2R (L) is connected to the upper body 24 via the hip joints (hip joints) 10R (L), 12R (L), and 14R (L).

  Each of the joints described above may have a known structure. In this case, the actuator that rotationally drives each joint is configured by, for example, an electric motor 42 (see FIG. 2) that includes a speed reducer.

  In this embodiment, between the ankle joints 18R (L), 20R (L) and the foot 22R (L) of each leg 2R (L), an elastic body such as rubber or a spring is omitted. Is inserted. Further, a sole elastic body 34 made of rubber or the like is attached to the bottom surface of each foot 22R (L). The spring mechanism 32 and the sole elastic body 34 function as a buffer mechanism or a compliance mechanism by being elastically deformed by the floor reaction force received by each leg 2R (L).

  With the above-described configuration of each leg 2R (L), the foot 22R (L) of each leg 2R (L) has six degrees of freedom with respect to the upper body 24. When the robot 1 moves, the joints of both legs 2R and 2L are 6 * 2 = 12 joints (in the present specification, “*” indicates multiplication in the calculation for the scalar and outer product in the calculation for the vector). By driving each at an appropriate angle, it is possible to perform a desired movement of both feet 22R and 22L. Thereby, the robot 1 can perform a movement that moves in a three-dimensional space, such as a walking action and a running action.

  Although illustration is omitted, in the present embodiment, a pair of left and right arms are attached to both sides of the upper portion of the upper body 24, and a head is mounted on the upper end portion of the upper body 24. Each arm body can perform a motion such as swinging the arm body back and forth with respect to the upper body 24 by a plurality of joints (shoulder joint, elbow joint, wrist joint, etc.) provided therein. . However, these arms and heads may be omitted.

  A control unit 26 that controls the operation of the robot 1 is stored in the upper body 24. In FIG. 1, for convenience of illustration, the control unit 26 is illustrated outside the upper body 24.

  Between the ankle joints 18R (L) and 20R (L) of each leg 2R (L) and the spring mechanism portion 32, there are six axes for measuring the floor reaction force acting on the foot 22R (L). A force sensor 36 is interposed. The six-axis force sensor 36 detects the translational force component in the three-axis direction and the moment component around the three axes of the floor reaction force transmitted from the foot 22R (L), and outputs the detection signal to the control unit 26. .

  The body 24 is equipped with an inclination sensor 38 for measuring the inclination angle (inclination angle in the roll direction and the pitch direction) of the body 24 with respect to the vertical direction (gravity direction) and the rate of change (angular velocity) thereof. . More specifically, the tilt sensor 38 includes an acceleration sensor and a rate sensor (angular velocity sensor) such as a gyro sensor, and outputs detection signals of these sensors to the control unit 26. Then, in the control unit 26, based on the output of the tilt sensor 38, the tilt angle and the angular velocity of the body 24 with respect to the vertical direction are measured by a known method.

  In addition, an electric motor 42 (see FIG. 2) that rotationally drives each joint is provided with an encoder (rotary encoder) 40 (see FIG. 2) for detecting the rotation angle of each joint. Is output to the control unit 26.

  Referring to FIG. 2, the control unit 26 is composed of an electronic circuit unit having a microcomputer, and includes a first arithmetic unit 50 and a second arithmetic unit 52 comprising an CPU, an A / D converter 54, a counter 56, D / A converter 58, RAM 60, ROM 62, and bus line 64 for exchanging data between them.

  In the control unit 26, the outputs of the six-axis force sensor 36 and the tilt sensor 38 are converted to digital values by the A / D converter 54 and then input to the RAM 60 via the bus line 64. Further, the output of the encoder (rotary encoder) 40 of each joint of the robot 1 is input to the RAM 60 via the counter 56.

  The first arithmetic unit 50 generates a target gait described later, calculates a joint displacement command (target value of the rotation angle of each joint), and sends it to the RAM 60. Further, the second arithmetic unit 52 reads out the joint displacement command from the RAM 60 and the actual joint displacement (actual value of the rotation angle of each joint) measured from the output of the encoder 40 via the counter 56, and the actual joint displacement. To drive the electric motor 42 of each joint necessary to follow the joint displacement command (command value defining the output torque of the electric motor 42).

  Then, the second arithmetic unit 52 outputs the calculated drive command to the servo amplifier 44 for driving the electric motor 42 via the D / A converter 58. At this time, the servo amplifier 44 drives the electric motor 42 according to the input drive command (energizes the electric motor 42). Thereby, the actual joint displacement of each joint is controlled to follow the joint displacement command.

  Next, the process executed by the control unit 26 of the robot 1 in this embodiment will be described more specifically with reference to the block diagram of FIG. Portions other than the “real robot” in FIG. 3 are functions realized by processing executed by the control unit 26 (mainly processing of the first arithmetic device 50 and the second arithmetic device 52).

  In FIG. 3, for the sake of convenience, actual values (actual joint displacement, etc.) recognized by the control unit 26 from the outputs of the respective sensors mounted on the robot 1 are shown as being output from the actual robot 1. In the following description, the symbols R and L are omitted when it is not necessary to distinguish between the left and right legs 2R and 2L.

  The control unit 26 includes a gait generator 100 that generates and outputs a desired gait freely and in real time as will be described later. This gait generator 100 functions as a target gait generator or a target motion generator in the present invention.

  The target gait output by the gait generator 100 includes a target motion trajectory (target motion trajectory) that defines a time-series pattern of target values of displacement amounts (rotation angles) of the joints of the robot 1, and the robot 1. It consists of a target trajectory related to the external force to be applied.

  In this embodiment, the target motion trajectory is a target body position / posture trajectory (trajectory of the target position and target posture of the body 24) and a target foot position / posture trajectory (trajectory of the target position and target posture of each foot 22). And the target arm posture trajectory (the trajectory of the target posture of each arm body). The target trajectory related to the external force is composed of a target total floor reaction force center point (target ZMP) trajectory and a target total floor reaction force trajectory. In addition to the leg body 2 and the arm body, when a portion that is movable with respect to the upper body 24 is provided, the target position / posture trajectory of the movable portion is added to the target motion trajectory.

  Here, the terms related to the target gait will be supplemented. In the following description, “target” is often omitted when there is no risk of misunderstanding.

  The “trajectory” in the target gait means a temporal change pattern (time-series pattern of instantaneous values).

  The position and speed of the upper body 24 mean the position of a predetermined representative point of the upper body 24 (for example, the center point between the left and right hip joints) and the moving speed thereof. Similarly, the position and speed of each foot 22 mean the position of the representative point determined in advance of each foot 22 and its moving speed. In this embodiment, the representative point of each foot 22 is a point on the bottom surface of each foot 22, for example, a perpendicular line from the center of the ankle joint of each leg 2 to the bottom surface of each foot 22 intersects the bottom surface. Set to a point.

  “Position” means a spatial orientation. For example, the body posture is represented by the inclination angle (posture angle) of the upper body 24 in the roll direction (around the X axis) and the inclination angle (posture angle) of the upper body 24 in the pitch direction (around the Y axis), The foot posture is represented by a biaxial spatial azimuth angle fixedly set on each foot 22. In this specification, the body posture is sometimes referred to as a body posture angle. The body posture may include the rotation angle of the body 24 in the yaw direction (around the Z axis).

  In addition, the floor reaction force (the floor reaction force consisting of translational force and moment) acting on each foot 22 is called “each foot floor reaction force”, and all (two) feet 22R and 22L of the robot 1 The resultant force of “each foot floor reaction force” is called “whole floor reaction force”. However, in the following description, since each foot floor reaction force is hardly mentioned, “floor reaction force” is treated as the same as “total floor reaction force” unless otherwise specified.

  The target floor reaction force is generally expressed by an action point and a translational force and a moment acting on the action point. However, in this embodiment, for the sake of convenience of explanation, it is assumed that an action point is distinguished from a translational force and a moment acting on the action point, and the target floor reaction force does not include the action point. In the present embodiment, the target floor reaction force center point is set as the action point of the target floor reaction force. In this case, the moment component of the desired floor reaction force is zero except for the vertical component (moment component around the Z axis).

  In the following description, the vertical component and the horizontal component of the translational force component (translational floor reaction force) of the floor reaction force are referred to as a floor reaction force vertical component and a floor reaction force horizontal component, respectively. Also, the vertical component and horizontal component of the moment component (floor reaction force moment) of the floor reaction force are referred to as the floor reaction force moment vertical component and the floor reaction force moment horizontal component, respectively.

  In ZMP (Zero Moment Point), the horizontal component (moment component around the horizontal axis) of the moment that the resultant force of the inertial force generated by the motion of the robot 1 and the gravity acting on the robot 1 acts around that point becomes zero. It means a point on the floor. In a gait that satisfies the dynamic equilibrium condition, the ZMP and the floor reaction force central point coincide. For this reason, in this embodiment, the target floor reaction force central point is often referred to as a target ZMP.

  In the present embodiment, in the desired gait, the desired body position / posture, the desired foot position / posture, the desired total floor reaction force central point (target ZMP), and the desired total floor reaction force are determined based on the operating environment ( It is described in a support leg coordinate system, which will be described later, as a global coordinate system fixed to the floor of the outside world. The target arm posture is described as a relative posture with respect to the upper body 24.

  The gait generator 100 according to the present embodiment alternately performs the leaving and landing operations (a series of operations including the leaving operation and the subsequent landing operation) of the legs 2 by the two legs 2 and 2. Thus, a gait for moving the robot 1 (a walking gait that causes the robot 1 to perform a walking operation or a traveling gait that performs a traveling operation) and a one-leg hopping operation (with one leg 2 leaving the floor) Further, it is possible to generate a gait (hereinafter referred to as a one-leg hopping gait) that causes the robot 1 to perform a floor leaving / landing operation using only the other leg 2.

  Among these gaits, the method of generating the walking gait or the running gait is the same as that described in the embodiment of Patent Document 1 in the present embodiment. On the other hand, in the one-leg hopping gait generation process, the method of setting some parameters for gait generation is different from the walking gait or running gait generation process.

  An outline of the operation of the robot 1 (single leg hopping operation) in the one leg hopping gait will be described with reference to FIG. In the following description, “landing” may be referred to as “landing” or “grounding”.

  The state at time t1 in FIG. 4 shows the initial state of the one-leg hopping gait. In this initial state, one leg 2 (the left leg 2L in the illustrated example) of the two legs 2, 2 is removed from the floor, and the other leg 2 (the right leg 2R in the illustrated example) is removed. It is in the state where it is made to land. That is, this initial state is a state where only one leg 2 is landing.

  The initial state may be a state in which the robot 1 has landed both legs 2 and 2 and stopped moving, but one of the legs 2 has left the floor, but is in the middle of walking or running. It may be.

  In the one-leg hopping operation, the robot 1 then kicks the floor with the other leg 2 (2R) and floats in the air with the one leg 2 (2L) leaving the floor (time t2). , Move in the air towards the target position.

  Next, the robot 1 lands on the other leg 2 (2R) (time t3). In this case, immediately after the landing, in order to reduce the landing impact, the robot 1 bends the other leg 2 (2R) at the knee joint while leaving the one leg 2 (2L) off the floor. As a result, the entire center of gravity of the robot 1 is lowered.

  Thereafter, when the one-leg hopping operation is continued, the same operation as the operation from time t1 is repeated. In addition, when the single leg hopping operation is shifted to, for example, the walking operation or the running operation, the one leg 2 (2L) that has left the floor at time t4 is landed.

  The above is the outline of the single leg hopping operation. In FIG. 4, the one-leg hopping operation for maintaining the left leg 2 </ b> L in the leaving state is illustrated, but the same applies to the one-leg hopping operation for maintaining the right leg 2 </ b> R in the leaving state.

  In the following description, in any of the walking gait, running gait and single-leg hopping gait of the robot 1, the leg 2 which supports the weight of the robot 1 out of the two legs 2 and 2 ( The leg 2) that receives the floor reaction force that supports its own weight may be referred to as a supporting leg or a supporting leg side leg, and the leg 2 that is not a supporting leg may be referred to as a free leg or a free leg side leg. In this case, in a running gait or a one-leg hopping gait having an aerial period in which both legs 2 and 2 are in the state of getting out of bed, in the aerial period, both legs 2 and 2 lose their own weight. Although it is not a supporting leg, for the sake of convenience, the leg 2 that was a supporting leg immediately before the aerial period is also referred to as a supporting leg or a supporting leg side leg even in the aerial period.

  In other words, the one-leg hopping gait is a gait having an aerial period and maintaining the supporting leg and the free leg in the same leg 2 before and after the aerial period. On the other hand, the walking gait or the running gait is, in other words, a gait in which the robot 1 is moved while the support leg and the free leg are alternately switched by the two legs 2 and 2 respectively.

  The support leg coordinate system will be described. In the present embodiment, the support leg coordinate system indicates the position and orientation of the foot 22 of the leg 2 that is landing as a support leg. A state in which almost the entire bottom surface of the base plate is in a position parallel to the floor surface and matched with the position / posture in contact (contact) with the floor surface (hereinafter, the position / posture of the foot 22 of the support leg in this state is the basic landing position / posture) The coordinate system fixed to the floor, in which the perpendicular line extending from the center of the ankle of the supporting leg side leg 2 to the floor surface intersects with the floor surface, and the horizontal plane passing through the origin is the XY plane It is. In this case, the X-axis direction and the Y-axis direction of the support leg coordinate system are respectively the front-rear direction and the left-right direction of the foot 22 of the support leg-side leg 2 (specifically, the foot 22 viewed on the horizontal plane). . The Z-axis direction is the vertical direction.

  Therefore, the support leg coordinate system in the present embodiment is such that the position of the origin and the directions of the X axis and the Y axis are relative to the basic landing position / posture of the foot 22 of the leg 2 that is landing as a support leg. It is a three-axis orthogonal coordinate system (XYZ coordinate system) defined according to the above relationship.

  It should be noted that the desired foot position / posture during landing of the support leg does not necessarily need to coincide with the basic landing position / posture at a certain time during the landing. In this case, the basic landing position / posture that defines the support leg coordinate system can be achieved by rotating the foot 22 landing in the target gait so as not to slip virtually. (Position and posture in which almost the entire bottom surface of the foot 22 is in contact with the floor surface).

  Supplementally, the origin of the support leg coordinate system does not necessarily have to be the above point, and may be a point on the floor surface shifted from the above point. In this embodiment, the support leg coordinate system is updated every time the support leg lands. However, the position of the origin of the support leg coordinate system and the attitude of the coordinate axes may be kept constant.

  Next, a dynamic model of the robot 1 used for gait generation processing in this embodiment will be described with reference to FIG. Since this dynamic model is the same as the dynamic model shown in FIG. 10 of Patent Document 1, only a schematic description is given in this embodiment.

  With reference to FIG. 5, the dynamic model of the robot 1 used for the gait generation process in the present embodiment is the same as the dynamic model shown in FIG. That is, this dynamic model has a total of three mass points consisting of two mass points 2m, 2m corresponding to each leg 2 of the robot 1 and a mass point 24m corresponding to the upper body 24, and inertia (moment of inertia). This model is composed of a flywheel FH having no mass.

  The mass of the upper mass point 24m includes the mass of a part (such as an arm body) that is connected to or fixed to the upper body 24 other than the legs 2 and 2.

  In this dynamic model, the dynamics of the legs 2 and 2 (the dynamics of the mass points 2m and 2m) and the dynamics of the upper body 24 (the dynamics of the mass point 24m and the flywheel FH) are configured to be non-interfering with each other. At the same time, the dynamics of the entire robot 1 is expressed by their linear combination.

  The behavior of this dynamic model (the dynamics of the robot on the dynamic model) is expressed as follows. However, to simplify the explanation, only the equation of motion in the sagittal plane (XZ plane (sagittal plane) including the X axis and Z axis of the support leg coordinate system) is described, and the lateral plane (support leg coordinate system The equation of motion in the YZ plane (frontal plane) including the Y axis and the Z axis is omitted.

  For convenience of explanation, variables and parameters relating to the dynamic model are defined as follows.

Zsup: Support leg mass point vertical position Zswg: Swing leg mass point vertical position Zb: Upper body mass point vertical position ZGtotal: Whole center of gravity vertical position Xsup: Support leg mass point horizontal position Xswg: Swing leg mass point horizontal position Xb: Upper body mass point horizontal position XGtotal: Whole center of gravity horizontal position θby: Body posture angle (tilt angle) around Y axis relative to vertical direction
mb: Mass of upper body mass
msup: Mass of support leg mass
mswg: Free leg mass
mtotal: Total robot mass (= mb + msup + mswg)
J: Moment of inertia of flywheel FH
Fx: Floor reaction force horizontal component
Fz: Floor reaction force vertical component
My: Floor reaction force moment around the target ZMP (specifically, the component around the Y axis of the floor reaction force moment)

In the present embodiment, the position of the mass point 2m of each leg 2 is a position defined according to the position and posture of the foot 22 of the leg 2. The position of the body mass point 24m is a position defined according to the position and posture of the body 24. Further, the rotation angle of the flywheel FH is assumed to coincide with the body posture angle.

  For any variable A, dA / dt represents the first derivative of A, and d2A / dt2 represents the second derivative of A. Therefore, if the variable A is displacement, dA / dt means velocity and d2A / dt2 means acceleration. G represents a gravitational acceleration constant. Here, g is a positive value.

  The equation of motion of the dynamic model (the dynamics of the robot on the dynamic model) is expressed by the following equation 01, equation 02, and equation 03.

Fz = mb * (g + d2Zb / dt2) + msup * (g + d2Zsup / dt2)
+ Mswg * (g + d2Zswg / dt2) ...... Formula 01
Fx = mb * d2Xb / dt2 + msup * d2Xsup / dt2 + mswg * d2Xswg / dt2
…… Formula 02
My = -mb * (Xb-Xzmp) * (g + d2Zb / dt2) + mb * (Zb-Zzmp) * d2Xb / dt2
-Msup * (Xsup-Xzmp) * (g + d2Zsup / dt2)
+ Msup * (Zsup-Zzmp) * d2Xsup / dt2
-Mswg * (Xswg-Xzmp) * (g + d2Zswg / dt2)
+ Mswg * (Zswg−Zzmp) * d2Xswg / dt2 + J * d2θby / dt2
...... Formula 03

Further, the following relational expression is established at the position of the entire center of gravity of the robot 1.
ZGtotal = (mb * Zb + msup * Zsup + mswg * Zswg) / mtotal …… Formula 04
XGtotal = (mb * Xb + msup * Xsup + mswg * Xswg) / mtotal ...... Formula 05

The above is the dynamic model used in the gait generation process of the gait generator 100. In this dynamic model, the motion of changing the body posture angle of the robot 1 is a flywheel so as not to move the entire center of gravity of the robot 1 (that is, the translational force reaction force acting on the robot 1 does not change). Expressed as FH rotational motion. The translational motion of the upper body 24 of the robot 1 is expressed as a translational motion of the upper body point 24b.

  Supplementally, the dynamic model may be different from that shown in FIG. For example, a dynamic model that expresses the dynamics of the robot 1 may be used by a multi-mass model having a mass at each link of the robot 1.

  Next, an outline of processing of the gait generator 100 will be described. The gait generator 100 includes data for specifying the type of motion of the robot 1 in the target gait to be generated (in this embodiment, walking motion, running motion, one-leg hopping motion), and the supporting leg-side leg 2. The data and the like for designating the position and orientation and timing of the foot 22 of the foot when landing are input. Then, the gait generator 100 sequentially generates a target gait for one step in units of the gait for one step based on these input data.

  Here, in the present embodiment, “a gait for one step” means that in the walking gait or the running gait, the landing operation of one leg 2 and the subsequent leg 2 are continued. It means the gait for the period between landing operations. In the one-leg hopping gait, “a gait for one step” refers to a landing operation of one leg 2 as a supporting leg and a next landing operation of the one leg 2. It means the gait for the period between.

  In the following explanation, the target gait to be newly generated (target gait for one step) is “current gait”, the next target gait is “next gait”, and the next target gait. Is called “next gait”. A target gait generated immediately before the “current time gait” is referred to as a “previous gait”.

  In the current time gait generation process, the gait generating device 100 uses the input data to generate a part of the current time gait such as a desired foot position / posture trajectory, a desired floor reaction force vertical component trajectory, and a desired ZMP trajectory. This time gait parameter is determined as a parameter that defines the constituent elements of.

  In this case, the gait generator 100 generates a virtual periodic gait following the current gait (the motion of the robot 1 having the same pattern) to generate the current gait that can ensure the continuity of the movement of the robot 1. Gait parameters are determined so that the current time's gait can be converged to the normal turning gait in the future. The gait generator 100 sequentially generates instantaneous values of the current time gait using the current time gait parameters and the dynamic model.

  Details of the gait generating process of the gait generating device 100 will be described below with reference to FIGS.

  The gait generator 100 generates a target gait by executing the processing shown in the flowchart of FIG. In the following description, detailed description of the same processing as that described in the embodiment of Patent Document 1 may be omitted.

  First, various initialization operations such as initializing time t to “0” in S010 are performed. This process is performed when the gait generator 100 is activated.

  Next, the process proceeds to S014 through S012, and the gait generator 100 waits for a timer interrupt for each control cycle (the calculation processing cycle in the flowchart of FIG. 6). The control period is Δt.

  Next, in S016, the gait generator 100 determines whether or not it is a gait change. At this time, if the gait is switched, the process proceeds to S020 through the process of S018. In this case, in S018, the gait generator 100 initializes the time t to “0”. If the gait is not switched at S016, the process proceeds to S032.

  Here, the “gait changeover” means the timing when generation of the previous time's gait is completed and generation of the current time's gait is started.

  In S020 next to S018, the gait generator 100 determines the operation mode of the robot 1 in the current time gait and the next time gait, and the support leg coordinate system of the next time and next time gait (specifically, the support leg coordinates). The position of the origin of the system and the posture of each coordinate axis) and the gait cycle of the current time gait and the next time gait.

  Here, the operation mode of the robot 1 is data designating the type of operation of the robot 1, and in this embodiment, the operation mode is any one of the walking operation mode, the traveling operation mode, and the one-leg hopping operation mode.

  In the next time's gait support leg coordinate system, the position of the origin and the direction (posture) of the coordinate axis are the planned landing position / posture of the foot 22 of the leg 2 that is planned as the support leg in the next gait (details). Is a support leg coordinate system defined by the above-described relationship according to the basic landing position / posture corresponding to a certain time during landing of the foot 22 (for example, a planned landing position / posture at the start of landing). . That is, the next time's gait supporting leg coordinate system has the basic landing corresponding to the planned landing position / posture of the foot 22 at the time of landing of the foot 22 of the supporting leg side leg 2 in the next time's gait. In a state where it matches the position and orientation, the origin is a point where a perpendicular extending from the center of the ankle joint of the supporting leg side leg 2 to the bottom surface of the foot 22 intersects the floor surface, and the foot 22 This is a support leg coordinate system in which the horizontal axis in the front-rear direction and the horizontal axis in the left-right direction are the X axis and the Y axis, respectively, and the vertical axis is the Z axis.

  Similarly, in the next gait support leg coordinate system, the position of the origin and the direction of the coordinate axis correspond to the planned landing position / posture of the foot 22 of the leg 2 which is planned as the support leg in the next gait. It is a support leg coordinate system prescribed | regulated by the relationship mentioned above according to the said basic landing position attitude | position.

  The current time gait cycle is the time from the start to the end of the current time gait, and the next time gait cycle is the time from the start to the end of the next time gait.

  In the present embodiment, the operation mode, the supporting leg coordinate system, and the gait cycle read by the gait generator 100 in S20 are the robot 1 given to the control unit 26 from a control device or server outside the robot 1 in this embodiment. Instruction information related to the operation (for example, information indicating the type of movement of the robot 1, movement speed, movement direction, movement plan, etc.), and environmental information related to the operation environment (external world) of the robot 1 (eg, floor shape information, map information, etc.) These are determined by a processing device (not shown) of the control unit 26 based on sensing information and the like recognized based on the output of the sensor group including the tilt sensor 38 and the like.

  Instead of the next gait supporting leg coordinate system and the next gait supporting leg coordinate system, the required value of the basic landing position / posture of the foot 22 of the supporting leg in each of the next gait and the next gait is determined. It may be determined by the control unit 26, and these required values may be read in S20.

  FIG. 7 shows the next time gait support leg coordinate system and the next time gait support leg coordinate system read in S20 when the operation mode of the current time gait and the next time gait is the operation mode for the single leg hopping operation, for example. Illustrated. In this example, the one-leg hopping gait in the current time gait and the next time gait is a gait with the right leg 2R as the supporting leg, and the robot 1 is roughly in the air during each of the current time gait and the next time gait. It is moving to the right. In the illustrated example, the previous time gait is a walking gait. Further, the foot 22 in FIG. 7 indicates the foot 22 in the basic landing position / posture corresponding to each support leg coordinate system.

  Next, proceeding to S022, the gait generator 100 determines the normal turning gait as a virtual periodic gait following the current gait (a virtual periodic gait that is a future convergence target of the current gait). Determine gait parameters. In this embodiment, the gait parameters include a foot trajectory parameter that defines a desired foot position / posture trajectory in a normal turning gait, a reference body posture trajectory parameter that defines a reference body posture trajectory, and a target arm posture. Specifies the arm posture trajectory parameter that defines the trajectory, the ZMP trajectory parameter that defines the target ZMP trajectory, the floor reaction force vertical component trajectory parameter that defines the target floor reaction force vertical component trajectory, and the allowable range of the target floor reaction force horizontal component Parameters.

  Here, the “steady turning gait” is the gait boundary (gait boundary for each cycle) when the gait is repeated as described in the embodiment of Patent Document 1. It means a periodic gait that does not cause discontinuity in the motion state of the robot 1 (the motion state such as the position and orientation of each part of the robot 1 and its changing speed). In other words, the “steady turning gait” is a periodic gait that can repeat the motion of the same pattern without causing discontinuity of the trajectory of the gait.

  In this embodiment, the normal turning gait that is a periodic gait is a gait for two steps of the robot 1, that is, a first turning gait following the current time gait and a second turning following the first turning gait. A gait consisting of gaits is a gait for one cycle, and the gait for one cycle is repeated at a constant cycle.

  Supplementing the normal turning gait (hereinafter sometimes referred to simply as the normal gait), the gait in which the support leg and the free leg are alternately switched between the two legs, such as the walking gait or the running gait, is steady. A gait for one cycle of the gait needs to include a gait for at least two steps. On the other hand, in a one-leg hopping gait in which the support leg and the free leg are held by the same leg 2, the gait for one cycle of the normal gait is the gait for one step. Good. However, in the present embodiment, in order to match the one-leg hopping gait generation process with the walking gait or the running gait generation process, even in the one-leg hopping gait, a step for one cycle of the normal gait. The gait is the gait for two steps. In this case, in this embodiment, in the normal gait of the one-leg hopping gait, the first turning gait and the second turning gait (more precisely, the first gait when viewed from the next time gait supporting leg coordinate system). The first turning gait and the second turning gait as viewed from the next time gait supporting leg coordinate system) are gaits having the same pattern.

  In this embodiment, even when generating a one-leg hopping gait, as in the embodiment of Patent Document 1, the following is used as an index for generating a target gait that enables continuous movement of the robot 1. The divergent component defined by Equation 10 is used. Note that ω0 is a predetermined value for a one-leg hopping gait.

Divergent component = Upper body point horizontal position + Upper body point horizontal velocity / ω0 ...... Equation 10

Then, the gait generator 100 determines the initial divergent component of the normal gait (the value of the divergent component at the start time of the normal gait), and adds the final divergent component of the current time gait to this initial divergent component. The current time gait is generated with the guideline of matching (the value of the divergent component at the end time of the current time gait). This generates a current time gait that can be converged to a normal gait in the future.

  Returning to the main subject, in S022, the gait generator 100 executes the subroutine processing shown in the flowchart of FIG.

  First, in S100, the gait generator 100 causes the foot in the gait parameters of the normal gait so that the foot position / posture trajectory is connected in the order of the current time gait, the first turning gait, and the second turning gait. Determine flat trajectory parameters.

  Here, when the next time's gait is a one-leg hopping gait as shown in FIG. 7, the normal gait is also a one-leg hopping gait. In this case, for example, the foot trajectory parameters are determined as follows. In the following description, the foot 22 of the supporting leg side leg 2 is referred to as a supporting leg foot 22, and the foot 22 of the free leg side leg 2 is referred to as a free leg foot 22. The “initial” and “end” of the gait mean the gait start time, end time, or instantaneous gait at those times, respectively.

  Foot trajectory parameters are composed of the position and orientation of the supporting leg foot and the free leg foot at the beginning and the end of the first turning gait and the second turning gait, the gait cycle of each turning gait, etc. Is done.

  If the current time's gait is a one-leg hopping gait, among the foot trajectory parameters, the first swing gait initial free leg foot position / posture is the current time's gait end play as viewed from the next time's gait supporting leg coordinate system. The leg foot position / posture is assumed.

  In the present embodiment, the foot position / posture at the end of the current time's gait viewed from the next time's gait supporting leg coordinate system is a predetermined position / posture (position / posture existing in the air. Hereinafter, one-leg hopping gait) This is referred to as the final free leg foot position / posture).

  Further, the initial support leg foot position / posture of the first turning gait is the support leg foot position / posture at the end of the current time gait viewed from the next time's gait support leg coordinate system.

  In this case, in this embodiment, when the current time gait is a one-leg hopping gait, the current time gait end support leg foot position / posture viewed from the next time gait support leg coordinate system is the next time gait support leg coordinate system. The basic landing position / posture (basic landing position / posture that defines the position of the origin of the next time's gait support leg coordinates and the directions of the X-axis and the Y-axis). That is, the current leg gait support leg foot position / posture is determined when the position / posture of the support leg foot 22 matches the current leg gait end support leg foot position / posture. The point where the perpendicular extending from the center of the ankle 22 to the floor intersects with the floor is the same as the origin of the next time's gait supporting leg coordinate system, and the longitudinal direction of the supporting leg foot 22 projected onto the horizontal plane And the left-right direction are basic landing position / postures that correspond to the X-axis direction and the Y-axis direction of the next time's gait support leg coordinate system, respectively.

  The free leg foot position / posture at the end of the first turning gait is expressed by the foot position / posture viewed from the next time's gait supporting leg coordinate system, and the foot position / posture is viewed from the next time's gait supporting leg coordinate system. When converted into a posture, the position is set so as to coincide with the free leg foot position / posture at the end of the one-leg hopping gait.

  The foot position / posture at the end of the first turning gait is expressed by the foot position / posture viewed from the next time's gait supporting leg coordinate system, and the foot position / posture is viewed from the next time's gait supporting leg coordinate system. The basic landing position / posture determined in accordance with the next gait support leg coordinate system (the position of the origin of the next gait support leg coordinate system and the directions of the X-axis and the Y-axis are defined when converted into a posture. The basic landing position / posture).

  The free leg foot position / posture at the initial stage of the second turning gait is the free leg foot position / posture at the end of the first turning gait as viewed from the next time's gait supporting leg coordinate system (that is, the free leg foot at the end of the one leg hopping gait) Flat position).

  The initial support leg foot position / posture of the second turning gait is the support foot / posture position / posture of the end of the first turning gait as viewed from the next gait support leg coordinate system (that is, the origin of the next gait support leg coordinate system). And the basic landing position / posture that defines the X-axis and Y-axis orientations).

  The free leg foot position / posture at the end of the second turning gait is represented by the foot position / posture as viewed from the next time's gait supporting leg coordinate system, and the first turning gait end position game as viewed from the next time's gait supporting leg coordinate system. It is set to match the leg foot position / posture.

  The foot position / posture at the end of the second turning gait supporting leg is expressed by the foot position / posture as viewed from the next time's gait supporting leg coordinate system, and is supported at the end of the first turning gait as viewed from the next time's gait supporting leg coordinate system. It is set to match the leg foot position / posture.

  The gait cycles of the first turning gait and the second turning gait are set to be the same as the next time gait cycle.

  The above is an example of the foot trajectory parameters determined in S100 when the next time's gait is a one-leg hopping gait.

  Next, proceeding to S102, a reference body posture trajectory parameter that defines a reference body posture trajectory that the target body posture should follow is determined.

  In the present embodiment, the reference body posture angle in a normal gait that repeats a one-leg hopping gait (hereinafter sometimes referred to as a single-leg hopping normal gait) is a constant value, and the reference body posture angle of this constant value is used. Is determined as the reference body posture trajectory parameter. The reference body posture angle is a posture angle (tilt angle) set as described below.

  Of the reference body posture angle of the one-leg hopping normal gait, the component around the Y axis (pitch direction) (inclination angle around the Y axis of the body 24 with respect to the vertical direction) is the XZ plane (sagittal) in this embodiment. It is set to a posture angle (a body posture angle when the robot 1 is erected in an upright posture) in which the direction of the trunk axis of the upper body 24 of the robot 1 viewed in a plane is a vertical direction.

  On the other hand, of the reference body posture angle of the normal gait for one leg hopping, the component around the X axis (roll direction) (the inclination angle of the body 24 around the X axis with respect to the vertical direction) is the YZ plane (lateral plane). The direction of the trunk axis of the upper body 24 of the viewed robot 1 is set to a posture angle that is inclined from the vertical direction to the right side or the left side of the robot 1.

  Specifically, as shown in FIG. 9, when the supporting leg of the one-leg hopping normal gait is the right leg 2R, the reference body posture angle around the X axis is the one-leg hopping normal gait. In the whole period of one cycle, the direction of the trunk axis of the upper body 24 is a predetermined angle in the positive direction around the X axis from the vertical direction (the clockwise direction toward the front of the robot 1). It is set to a posture angle that is inclined by the body posture angle in the roll direction for leg hopping.

  When the supporting leg of the one-leg hopping normal gait is the left leg 2L, the reference body posture angle around the X axis is the upper body in the entire period of one cycle of the one-leg hopping normal gait. 24 trunk axes tilted in the negative direction (counterclockwise direction toward the front of the robot 1) in the roll direction (around the X axis) from the vertical direction by the body posture angle in the roll direction for one leg hopping. It is set to the posture angle that becomes the thing.

  In the description of the present embodiment, the state of the trunk axis of the body 24 matches the vertical direction (the body posture matches the body posture when the robot 1 stands upright). The reference body posture angle at is set to “0”.

  The one-leg hopping roll direction body posture angle is set with a pointer as described below with reference to FIGS. 10 and 11.

  FIG. 10 shows that the body posture angle is set to “0” without tilting the upper body 24 in the landing state of the supporting leg (right leg 2R in the illustrated example) after the mid-air period in the one-leg hopping gait. The operation state of the robot 1 is shown. This operation state corresponds to an operation state at a time between times t3 and t4 in the one-leg hopping operation shown in FIG.

  In the state immediately after landing of the support leg after the mid-air period in the one-leg hopping gait, the action line of the translational floor reaction force acting on the robot 1 from the contact surface of the support leg foot 22 (22R) is shown in the figure. Thus, it is necessary to pass near the entire center of gravity Ga of the robot 1 (the center of gravity point Ga of the entire robot 1). This is because if the action line is far away from the overall center of gravity Ga, a moment due to the floor reaction force is generated around the overall center of gravity Ga, and the entire robot 1 rotates greatly.

  In this situation, as shown in FIG. 10, if the body posture angle is set to “0” without tilting the body 24, it is between the hip joint of the support leg and the line of action of the translational floor reaction force. The distance Ofs in the Y-axis direction (lateral direction of the robot 1) is relatively large.

  For this reason, a relatively large moment acts on the hip joint of the support leg in the roll direction (around the X axis) due to the substantially vertical translational floor reaction force. In particular, after the landing of the support leg after the mid-air period, the translational floor reaction near the time when the height of the overall center of gravity Ga of the robot 1 becomes the lowest by bending the support leg (near time t4 in FIG. 4). Since the force is increased, the moment is also increased. For this reason, it is necessary to apply a relatively large driving torque against the moment to the hip joint of the support leg.

  Therefore, in this embodiment, in order to reduce the moment acting on the hip joint of the support leg in the landing state of the support leg after the mid-air period, the upper body 24 in the one leg hopping gait is as shown in FIG. Inclined. In this case, more specifically, the tilted body posture is relatively more in the hip joint of the free leg (left leg 2L in the illustrated example) than in the hip joint of the support leg (right leg 2R in the illustrated example). The body posture is obtained by inclining the direction of the trunk axis of the body 24 from the vertical direction so as to be in a high position.

  When the support leg is the right leg 2R, the trunk axis of the upper body 24 is set in the vertical direction so that the hip joint of the free leg is positioned higher than the hip joint of the support leg. Inclining is realized by inclining the trunk axis of the upper body 24 with respect to the vertical direction in the positive direction around the X axis (clockwise toward the front of the robot 1). When the support leg is the left leg 2L, the trunk axis of the upper body 24 is inclined from the vertical direction so that the hip joint of the free leg is positioned relatively higher than the hip joint of the support leg. This is realized by inclining the trunk axis of the upper body 24 with respect to the vertical direction in the negative direction around the X axis (counterclockwise toward the front of the robot 1).

  In this way, by tilting the upper body posture of the robot 1 with respect to the vertical direction around the X axis, the overall center of gravity Ga of the robot 1 is located almost directly above the ground contact surface of the support leg foot 22. In the state where the joints of the support leg side leg 2 are operated as described above, the distance Ofs in the Y-axis direction (lateral direction of the robot 1) between the hip joint of the support leg and the action line of the translational floor reaction force is Compared to the case where the body posture angle is “0” (the case shown in FIG. 10), the posture angle can be reduced as shown in FIG. As a result, the moment acting on the hip joint of the support leg can be reduced by the translational floor reaction force.

  The one-leg hopping roll direction body posture angle shown in FIG. 9 is set so as to reduce the moment acting on the hip joint of the support leg as described above. Value. The body posture angle in the roll direction for one leg hopping is substantially equal to the body posture angle of the robot 1 and the ground contact surface of the supporting leg foot 22 on which the entire center of gravity Ga of the robot 1 is landed. When each joint of the supporting leg side leg 2 is moved so that it exists at a position directly above, the distance Ofs is set to “0” or a value within a predetermined range near “0”. Has been.

  In this embodiment, the reference body posture angle of the one-leg hopping normal gait is set to a constant value throughout the normal gait, but the reference body posture angle is set to one cycle of the one-leg hopping normal gait. You may make it change continuously within this period. However, in this case, within the period when the support leg is in the grounded state after the aerial period (particularly, within the period in the vicinity where the magnitude of the vertical component of the floor reaction force acting on the robot 1 is maximum), the X axis The reference body posture angle around the X axis is changed so that the surrounding reference body posture angle is equal to or close to the one-leg hopping roll direction body posture angle.

  In the following description, unless otherwise specified, it is assumed that the “normal gait” or “normal turning gait” is a one-leg hopping normal gait.

  Returning to the description of FIG. 8, after determining the reference body posture trajectory parameter (reference body posture angle) as described above, the process proceeds to S <b> 104, and the gait generator 100 determines the Z axis (or the body trunk). Arm posture trajectory parameters other than those related to changes in angular momentum of both arms around the axis) are determined. For example, a parameter that defines the trajectory of the relative center of gravity position of the entire arm body with respect to the upper body 24 is determined as the arm posture trajectory parameter. In this case, in the present embodiment, the relative center-of-gravity position of the entire arm body is set to be kept constant with respect to the upper body 24.

  Next, proceeding to S106, the gait generator 100 determines a floor reaction force vertical component trajectory parameter. Specifically, the floor reaction force vertical component trajectory of the one-leg hopping normal gait is set in a pattern as shown in FIG. 12, for example. In that pattern, in both the first turning gait and the second turning gait, the floor reaction force vertical component is trapezoidal during the one-leg support period (the period during which the support leg of the one-leg hopping gait is landing). The floor reaction force vertical component is maintained at “0” in the air. Then, the break point time of the pattern and the height (peak value) of the trapezoidal portion are determined as the floor reaction force vertical component trajectory parameters.

  In this floor reaction force vertical component trajectory parameter, the average value of the floor reaction force vertical component in the period of one cycle of the one-leg hopping normal gait is the same magnitude as the gravity acting on the entire robot 1 and is in the opposite direction. To be determined. This is the initial state of all state quantities of the normal gait (motion state quantities such as the position, posture, speed, etc. of each part of the robot 1) as seen from the supporting leg coordinate system of the first turning gait of the normal gait. The initial state of the first turning gait) and the terminal state (the terminal state of the second turning gait as seen from the supporting leg coordinate system of the first turning gait following the second turning gait of the normal gait) are the same. This is because it is necessary to satisfy a condition that it is satisfied (hereinafter, this condition may be referred to as a boundary condition of a normal gait).

  Next, in S108, the gait generator 100 determines the floor reaction force horizontal component permissible range [Fxmin, Fxmax] (more specifically, parameters defining this) according to the floor reaction force vertical component trajectory set as described above. ) Is set.

  The allowable ranges [Fxmin, Fxmax] are set as shown in FIG. 13, for example. The negative broken line in FIG. 13 represents the floor reaction force horizontal component allowable lower limit value Fxmin, and the positive broken line represents the floor reaction force horizontal component allowable upper limit value Fxmax.

  These floor reaction force horizontal component permissible lower limit value Fxmin and floor reaction force horizontal component permissible upper limit value Fxmax are the frictional force that can be generated between the support leg foot 22 and the floor surface in terms of their magnitude (absolute value). It is set so as not to exceed the upper limit of the size of.

  In the illustrated example, Fxmin and Fxmax are set from the following expressions 12a and 12b according to the floor reaction force vertical component determined in S106. Here, for convenience of explanation, the floor is assumed to be a horizontal plane.

Fxmin = −ka * μ * Floor reaction force vertical component …… Formula 12a
Fxmax = ka * μ * Floor reaction force vertical component ...... Formula 12b

In these equations 12a and 12b, ka is a positive constant (<1), and μ is a coefficient of friction between the foot 22 and the floor. In this case, as a parameter for defining the floor reaction force horizontal component allowable range, in the example, the value of (ka * μ) in Expressions 12a and 12b can be used.

  The floor reaction force horizontal component allowable range may be set by another method as long as it is within the limit of the frictional force that can be generated between the foot 22 and the floor surface.

  Next, proceeding to S110, the gait generator 100 determines a ZMP trajectory parameter that defines a target ZMP trajectory of the one-leg hopping normal gait. In this case, the ZMP trajectory parameter is determined so that the target ZMP defined thereby changes as follows, for example.

  That is, the target ZMP trajectory in the first turning gait of the one-leg hopping normal gait is, for example, the support leg foot 22 (22R in the illustrated example) corresponding to the next gait support leg coordinate system of FIG. It is determined as shown by the target ZMP trajectory with the support leg foot 22 (22R in the illustrated example) corresponding to the support leg coordinate system. Of the target ZMP trajectory shown in FIG. 7, the solid line indicates the target ZMP trajectory when the support leg foot 22 is landing, and the broken line indicates the target ZMP trajectory in the air phase. Is shown.

  In the example shown in FIG. 7, the substantially center point in the contact surface of the supporting leg foot 22 (22R in the illustrated example) corresponding to the next time gait supporting leg coordinate system is the initial (first turning) of the one-leg hopping normal gait. It is set as the target ZMP in the initial stage of the gait. In addition, a substantially center point within the contact surface of the support leg foot 22 (22R in the illustrated example) corresponding to the next time gait support leg coordinate system is set as a target ZMP at the end of the first turning gait. The target ZMP trajectory in the first turning gait is set such that the position of the ZMP continuously changes from the initial target ZMP to the final target ZMP. In this case, the target ZMP in the one-leg support period may be kept constant during a part of the one-leg support period (for example, a period immediately after the start of the one-leg support period).

  Although illustration is omitted, the target ZMP trajectory in the second turning gait of the one-leg hopping normal gait is also set in the same manner as described above.

  In this case, the target ZMP at the initial stage and the end of each of the first turning gait and the second turning gait, the target ZMP immediately before the start of the aerial period (immediately before the end of the one-leg support period), etc. Is determined as the ZMP trajectory parameter.

  The target ZMP in the one-leg support period is desirably set so as not to approach the boundary of the ground contact surface of the support leg foot 22 (so as to increase the so-called stability margin). The target ZMP is preferably set so as not to change rapidly.

  Next, in S112, the gait generator 100 redefines the initial time Ts and the end time Te of one cycle of the normal gait as follows.

  In this embodiment, in order to facilitate the determination of the normal gait, the initial time Ts and the end time Te of one step of the normal gait are determined as shown in FIG. 12 for the sake of convenience until the normal gait is determined. . That is, the time Ts when the floor reaction force vertical component is reduced to some extent in the second half of the one-leg support period of the first turning gait is set as the initial time Ts of the normal gait.

  Further, a time obtained by adding one cycle Tcyc of the normal gait (the total time of the first turning gait and the second turning gait) to the initial time Ts is set as the end time Te of the normal gait. .

  In this embodiment, after the normal gait is determined, the initial time of the normal gait is returned to the original time (the time when the support leg foot 22 landed). In the following description, the original initial time of the normal gait is set to “0”, and is distinguished from the initial time Ts used until the normal gait is determined.

  The above is the details of the processing in S022 of FIG.

  Returning to FIG. 6, the gait generator 100 executes the process of S022 as described above, and then proceeds to S024 to calculate the initial state of the normal gait. Specifically, the initial state calculated here is the initial body position velocity (initial body position and initial body velocity) of the normal gait, the initial divergence component, the initial body posture angle, and the angular velocity thereof. The calculation of these initial states is carried out exploratively by the processing shown in the flowchart of FIG.

  To describe below, first, in S200, the gait generator 100 determines the desired foot position / posture, the desired arm posture of the normal gait based on the normal gait parameters (gait parameters determined in S022 of FIG. 6), And the initial state (state at the initial time Ts) of the target body posture angle is determined. Here, the “state” means a set of a position or posture and a temporal change rate (change speed of the position or posture).

  In this case, the initial state of the foot position / posture on the support leg side and the initial state of the foot position / posture on the free leg side are determined based on the foot trajectory parameters determined in S100 of FIG.

  Specifically, with regard to the initial state of the foot position / posture on the support leg side in the one-leg hopping normal gait, the foot leg / posture at the end of the first turning gait from the first foot position / posture of the first turning gait initial supporting gait. The position / posture trajectory of the supporting leg foot 22 reaching the position / posture is generated by using a finite time settling filter in the same manner as the case of generating the foot position / posture trajectory in the running gait in the embodiment of Patent Document 1. . Then, the position and posture at the initial time Ts in the foot position / posture trajectory, and the respective change speeds (temporal change rates) of the position and posture are determined as the initial state of the foot position / posture on the supporting leg side. Is done.

  Similarly, regarding the initial state of the foot position / posture on the free leg side in the one-leg hopping normal gait, the first leg position / posture at the end of the first turning gait from the initial foot position / posture of the first turning gait. The position / orientation trajectory of the free leg foot 22 is generated using a finite time settling filter. Then, the position and posture at the initial time Ts in the foot position / posture trajectory and the changing speeds of the position and posture are determined as the initial state of the foot position / posture on the free leg side.

  In this case, the maximum value of the height (vertical position) of the supporting leg foot and the free leg foot in the mid-air period of the one-leg hopping normal gait is, for example, the first supporting leg leg in the first turning gait. The distance between the position of the supporting leg foot corresponding to the flat position and posture and the position of the supporting leg foot corresponding to the end supporting leg foot position and posture of the first turning gait, and the gait of the first turning gait. It is determined according to the period or the like.

  In addition, the initial state of the target arm posture is obtained based on the arm posture trajectory parameter determined in S104 of FIG. 8 and the arm posture at the time Ts (the overall center of gravity position of both arms with respect to the upper body 24) and the rate of change thereof. Is determined.

  In addition, the initial state of the target body posture angle is that the reference body posture angle and its angular velocity at the time Ts defined by the reference body posture trajectory parameter determined in S102 of FIG. Determined as a state. In this embodiment, since the reference body posture angle in the one-leg hopping normal gait is a constant value, the initial state of the angular velocity of the target body posture angle is “0”.

  Next, in S202, the gait generator 100 is an initial body horizontal position speed candidate (horizontal position and horizontal speed candidate of the body 24 at time Ts) (Xs, Vxs) (Xs: horizontal position, Vxs: (Horizontal speed) is provisionally determined. The candidates (Xs, Vxs) to be tentatively determined here may be arbitrary. For example, the body horizontal position speed in the initial state (state at time Ts) of the normal gait obtained at the time of generating the previous gait is a candidate (Xs , Vxs).

  In order to simplify the description, a case where an initial state (initial body horizontal position speed) of a normal gait in the X-axis direction is searched on the sagittal plane (XZ plane) will be described as an example. In practice, however, the initial state of the normal gait (the initial state satisfying the boundary condition of the normal gait) is searched separately for the position and the velocity separately in the X-axis direction and the Y-axis direction or simultaneously. There is a need.

  As an exploratory determination method, a pseudo Jacobian (sensitivity matrix) is obtained, and the next candidate is determined by the steepest descent method or the simplex method. In the present embodiment, for example, the steepest descent method is used.

  Next, the process proceeds to S206 via S204, and the gait generator 100 determines the vertical position (Z-axis direction position) Zs and the vertical speed (Z-axis direction speed) of the body 24 at the initial time (time Ts) of the normal gait. An initial body vertical position velocity (Zs, Vzs) which is a set of Vzs is determined.

  The initial body vertical velocity Vzs is determined in the present embodiment as follows, for example.

  The following formula is established for the robot 1 as a dynamic relationship.

Terminal total center-of-gravity vertical position-initial total center-of-gravity vertical position = (floor reaction force vertical component / total mass of robot) + second-order integral + gravity acceleration second-order integral + initial total center-of-gravity vertical velocity * 1 period time
... Formula 13

However, the gravitational acceleration of Equation 13 is a negative value.

  Further, in the normal gait, the terminal overall center-of-gravity vertical position and the initial overall center-of-gravity vertical position coincide with each other, so the right side of Equation 13 must be “0”. Therefore, the initial overall center-of-gravity vertical velocity can be obtained from these relationships.

  Specifically, first, a value obtained by dividing the floor reaction force vertical component calculated from the floor reaction force vertical component trajectory parameter set in S106 of FIG. 8 by the total mass of the robot 1 is obtained as one cycle of the normal gait. By performing second-order integration over a period (period from time Ts to time Te), the total center-of-gravity movement amount (first term on the right side of Equation 13) by the floor reaction force vertical component is obtained.

  Furthermore, the total center-of-gravity movement amount due to gravity (the second term on the right side of Equation 13) is obtained by integrating the gravitational acceleration on the second floor in the period of one cycle of the normal gait. Then, by inverting the sign of the sum of the total center-of-gravity movement amount due to the floor reaction force and the total gravity center movement amount due to gravity as described above, dividing by the time of one cycle Tcyc of the normal gait, the initial total The center-of-gravity vertical velocity is determined.

  Next, in order to obtain the initial body vertical position Zs, the overall center-of-gravity vertical velocity at time “0” is obtained using the following equation (14). In this case, the value obtained as described above based on Equation 13 is substituted for the overall center-of-gravity vertical velocity at time Ts in Equation 14 below. The integration period is a period from time “0” to time Ts.

Total center-of-gravity vertical velocity at time Ts−total center-of-gravity vertical velocity at time “0” = (first floor integral of floor reaction force vertical component / total mass of robot) + first-order integral of gravity acceleration
... Formula 14
(However, gravity acceleration is a negative value.)

Next, for example, the body vertical position at time “0” is determined using the body height determination method previously proposed by the applicant of the present application in Japanese Patent Laid-Open No. 10-86080. At this time, the foot position / posture at the time “0” (the first foot position / posture of the first turning gait initial supporting gait determined in S100 of FIG. 8 and the foot position / posture of the first swing gait initial free leg) and each leg Based on a predetermined geometric condition regarding the bending angle of the knee of the body 2, the body vertical position (body height) at which the knee of each leg 2, 2 does not fully extend at time “0”. Is determined.

  Specifically, when the knee bending angle of the supporting leg side leg 2 is θsup and the knee bending angle of the free leg side leg 2 is θswg, for example, the sum of the reciprocals of the sine values of the knee bending angles θsup and θswg The body vertical position is determined so that becomes a predetermined value (finite value). Here, the knee bending angles θsup and θswg are angles of the axis of the lower leg with respect to the axis of the thigh of each leg 2, and the knee is bent from the state where each leg 2 is fully extended. Accordingly, the angle increases from “0”.

  Next, the total center-of-gravity vertical position at time “0” is obtained from the vertical body position at time “0” and the foot position / posture determined as described above. For example, by using the kinematic model of the equation 04 corresponding to the model shown in FIG. 5, the overall center-of-gravity vertical position at time “0” is obtained.

  Specifically, the vertical position of the body mass point 24m is obtained from the body vertical position and body posture angle at time “0”. Also, the supporting leg foot position / posture and the free leg foot position / posture at time “0” (these are the first supporting gait initial position / posture and the first turning gait set in S100 of FIG. 8). The vertical positions of the supporting leg mass point 2m and the free leg mass point 2m are obtained respectively. Then, by substituting the obtained vertical positions of the upper body mass point 24m, the supporting leg mass point 2m, and the free leg mass point 2m into Zb, Zsup, and Zswg in the above-described equation 04, the total center-of-gravity vertical position (ZGtotal in the equation 04) Is required. It should be noted that a more rigorous model of the robot 1 (for example, a model having a mass point corresponding to each link of the robot 1) is used and the horizontal position and arm posture of the upper body at the time “0” are also taken into consideration. The total center-of-gravity vertical position at “” may be obtained.

  Next, in Formula 13, the overall center of gravity vertical position at time “0” obtained as described above is substituted into the initial overall center of gravity vertical position on the left side of Equation 13, and at time “0” obtained as described above. Substituting the total center-of-gravity vertical velocity into the initial total center-of-gravity vertical velocity on the right side of equation (13), and further assuming that one cycle time of equation (13) is Ts (= time from time “0” to time Ts), and the integration period is By setting the period from time “0” to time Ts, the value of the terminal overall center of gravity vertical position on the left side of Equation 13 is calculated, and this calculated value is obtained as the initial (time Ts) overall center of gravity vertical position.

  Further, from the obtained initial (time Ts) overall center-of-gravity vertical position and the foot position / posture at the time Ts (the foot position / posture in the initial state determined in S200), a model of the robot 1 (for example, kinematics of Formula 04). Model) is used to determine the initial (time Ts) body vertical position Zs.

  Specifically, the vertical positions of the support leg mass point 2m and the free leg mass point 2m of the model shown in FIG. 5 are obtained from the flat position / posture of the support leg and the free leg at time Ts. Then, by applying these vertical positions and the initial (time Ts) overall center-of-gravity vertical position obtained as described above to Expression 04, the vertical position of the body mass point 24m (Zb in Expression 04) is obtained. The initial (time Ts) body vertical position Zs is obtained from the vertical position of the body mass point 24m and the body posture at the time Ts (the body posture in the initial state determined in S200). In this case as well, the initial body vertical position Zs may be obtained using a more rigorous model and taking into account the body horizontal position and arm posture.

  Furthermore, from the initial state of the foot position / posture determined in S200 and the initial overall center-of-gravity vertical velocity determined in S206, the initial (time Ts) is determined using the model of the robot 1 (for example, the kinematics model of Expression 04). The body vertical velocity is required.

  Specifically, an equation obtained by differentiating both sides of the equation 04 with respect to time is defined by the initial overall center-of-gravity vertical velocity obtained based on the equation 13 and the initial state of each foot position / posture of the support leg and the free leg. The vertical velocity of the upper body mass point 22m is obtained by applying the vertical velocity of the supporting leg mass point 2m and the free leg mass point 2m. The body vertical velocity is determined from the vertical velocity of the body mass point 22m and the initial state of the body posture angle (determined in S200 or S208 described later). In addition to the initial state of each foot position / posture, the initial overall center-of-gravity vertical velocity, the initial state of the arm posture (determined in S200), and the initial state of the temporarily determined body horizontal position (S202 or S216 described later) Alternatively, the initial body vertical velocity that satisfies the initial overall center-of-gravity vertical velocity obtained as described above in consideration of the initial body vertical position obtained as described above is determined by the robot 1 in consideration of the initial body vertical position obtained as described above. You may make it obtain | require using a more exact model.

  After the process of S206, the process then proceeds to S208, where the gait generator 100 temporarily generates a gait as a normal gait candidate (a gait for one cycle of the normal gait). More specifically, based on the gait parameters of the normal gait determined in S022 of FIG. 6, the target ZMP and the target at each moment (time at each predetermined step time) from the initial time Ts to the end time Te. The floor reaction force vertical component, the desired foot position / posture, the reference body posture, the target arm posture, and the floor reaction force horizontal component allowable range are sequentially obtained.

  Then, using the dynamic model (model of FIG. 5) so as to satisfy the dynamic equilibrium condition regarding the obtained target ZMP and the desired floor reaction force vertical component and the condition of the floor reaction force horizontal component allowable range. By successively determining the body position and posture, a normal gait for one cycle from the initial time Ts to the end time Te is generated.

  In this case, the body horizontal position speed (Xs, Vxs) and the body vertical position speed (Zs, Vzs) are set to the initial state (time Ts) of the body 24. Further, the body posture is generated so as to coincide with the reference body posture as much as possible.

  Note that the generation of the normal gait in S208 is only performed inside the gait generator 100, and is not output from the gait generator 100.

  Specifically, the process of S208 is executed as shown in the flowchart of FIG.

  In the following, the gait generator 100 first performs various initializations in S300.

  Specifically, the initial time Ts of the normal gait is substituted for the gait generation time Tk. In addition, the latest candidate value of the initial (time Ts) body horizontal position speed (Xs, Vxs) (the latest candidate value determined in S202 of FIG. 14 or S216 or S218 described later) is set as the body horizontal position speed. The latest value of the initial (time Ts) body vertical position speed (Zs, Vzs) (the latest value determined in S206 of FIG. 14) is substituted for the body vertical position speed. Also, the initial value of the reference body posture angle (reference body posture at time Ts) is substituted into the body posture angle, and the initial value of the reference body posture angular velocity (reference value at time Ts) is substituted into the body posture angular velocity. (Body posture angular velocity) is substituted. Supplementally, as will be described later, the initial value of the body posture angular velocity is changed at the initial stage (time Ts), so the initial value of the body posture angular velocity is set to a value different from the initial value of the reference body posture angular velocity. May be.

  Next, in S304 via S302, the gait generator 100 determines whether the gait generation time Tk (current value) is a time before the end time Te (= Ts + Tcyc) (Tk ≦ Te). ). If this determination result is affirmative, the gait generator 100 determines the instantaneous value of the gait at time Tk by executing the process of S306 (details will be described later).

  Next, in S308, the gait generator 100 increases the gait generation time Tk by a predetermined step time ΔTk, and then performs the determination in S304 again. Here, the step time ΔTk may be set to coincide with the control cycle Δt, for example. However, in order to reduce the amount of calculation, ΔTk may be set to a time longer than Δt.

  If the determination result in S304 is negative, the process proceeds to S310. With the above processing, a normal gait is generated from the initial time (time Ts) to the end (time Te) before proceeding to S310.

  Details of the process for determining the instantaneous value of the normal gait in S306 will be described below with reference to the flowcharts of FIGS.

  The process of S306 is executed as shown in the flowchart of FIG.

  First, in S400, the gait generator 100 obtains an instantaneous value at the time Tk of the desired floor reaction force vertical component shown in FIG. 12 based on the normal gait parameter (specifically, the floor reaction force vertical component trajectory parameter). .

  Next, in S402, the gait generator 100 obtains an instantaneous value of the target ZMP at time Tk based on the normal gait parameters (specifically, ZMP trajectory parameters).

  Next, in S404, the gait generator 100, based on the normal gait parameters (specifically, the foot trajectory parameters, the reference body posture trajectory parameters, and the arm posture trajectory parameters), the desired foot position / posture at the time Tk, Instantaneous values of the reference body posture and the target arm posture are obtained. However, regarding the target arm posture, the position of the center of gravity of both arms is determined in more detail, but the movement of the arm (arm swing) that changes the angular momentum around the vertical axis (or the trunk axis of the upper body 24). Exercise) has not yet been determined.

  Note that the instantaneous value of the desired foot position / posture at time Tk is obtained using a finite time settling filter in the same manner as when the foot position / posture at initial time Ts is obtained in S200 of FIG.

  Next, in S406, the gait generator 100 satisfies the target floor reaction force vertical component (the total center of gravity such that the sum of the vertical inertia force and gravity of the robot 1 is balanced with the target floor reaction force vertical component). An instantaneous value (instantaneous value at time Tk) of the vertical position velocity is calculated.

  Specifically, by dividing the sum of the floor reaction force vertical component (≧ 0) and the gravity acting on the entire robot 1 (here, gravity <0) by the total mass of the robot 1, the robot 1 The overall center-of-gravity vertical acceleration (the vertical acceleration of the overall center of gravity) is obtained.

  Then, the total center-of-gravity vertical velocity at time Tk is calculated by integrating the total center-of-gravity vertical acceleration. Further, by integrating this total center-of-gravity vertical velocity, the total center-of-gravity vertical position at time Tk is calculated. More specifically, these calculations are performed by the following equations 15 and 16.

Total center-of-gravity vertical velocity at time Tk = total center-of-gravity vertical velocity at time (Tk−ΔTk) + ((reaction force vertical component at time Tk / total mass of robot))
+ Gravity acceleration) * ΔTk
...... Formula 15

Overall center of gravity vertical position at time Tk = Overall center of gravity vertical position at time (Tk−ΔTk) + Overall center of gravity vertical speed at time Tk * ΔTk
...... Formula 16

Note that the gravitational acceleration in Expression 15 is a negative value.

  Next, in S408, the gait generator 100 calculates a body vertical position that satisfies the overall center-of-gravity vertical position at time Tk. Specifically, for example, the body vertical position is calculated using the above-described equation 04 related to the model of FIG. That is, the vertical positions of the support leg mass point 2m and the free leg mass point 2m are obtained from the target foot position / posture of the support leg and the free leg at time Tk. Then, the vertical position of the upper mass point 24m is obtained by applying the vertical position of the support leg mass point 2m and the free leg mass point 2m and the total center-of-gravity vertical position at time Tk obtained in S406 to Expression 04. . Further, the body vertical position is obtained from the obtained vertical position of the body mass point 24m and the target body posture angle at time (Tk−ΔTk) (or the predicted value of the target body posture angle at time Tk).

  Note that the body vertical position may be obtained using a more strict model (for example, a model having a mass point corresponding to each link of the robot 1).

  Next, the process proceeds to S410, and the gait generator 100 determines the floor reaction force horizontal component allowable range [Fxmin shown in FIG. 13 based on the normal gait parameters (specifically, parameters that define the floor reaction force horizontal component allowable range). , Fxmax] at the time Tk.

  Next, in S412, the gait generator 100 determines the body horizontal acceleration and the body posture angular acceleration at time Tk. In this case, the body horizontal acceleration and the body posture angular velocity satisfy the target ZMP (specifically, the resultant force of the inertial force generated by the target motion of the robot 1 and the gravity acting on the robot 1 is about the target ZMP. In addition to satisfying the condition that the horizontal component of the moment generated in the robot is “0”), the floor reaction force horizontal component Fx that balances the horizontal inertial force of the robot 1 deviates from the allowable range [Fxmin, Fxmax]. And the condition that the body posture angular velocity coincides with the initial time Ts and the end time Te of the normal gait.

  The process of S412 is executed as shown in the flowchart of FIG.

  First, in S500, the gait generator 100 sets times Tm, Ts2, and Tm2 shown in FIG. In the present embodiment, the time Tm is set to a time immediately after the start of the second turning gait and the floor reaction force vertical component has increased to a predetermined magnitude. The time Ts2 is set to a time when the floor reaction force vertical component is reduced to some extent in the second half of the one-leg support period of the second turning gait. Also, the time Tm2 is set to a time immediately after the start of the first turning gait after the mid-air period of the second turning gait and when the floor reaction force vertical component has increased to a certain magnitude.

  Next, in S502, the gait generator 100 determines that the current time Tk is one of the time within the period from time Ts to Tm and the time within the period from time Ts2 to Tm2. Judge whether there is. If the determination result in S502 is affirmative (if the allowable range of the floor reaction force horizontal component at time Tk is relatively narrow or “0”), the process proceeds to S504. If the determination result in S502 is negative (if the allowable range of the floor reaction force horizontal component at time Tk is relatively wide), the process proceeds to S518.

  In step S504, the gait generator 100 determines that the body posture angular acceleration at time Tk is the reference body posture angular acceleration β0 at time Tk (the reference body posture trajectory parameter defined for the one leg hopping gait). The body horizontal acceleration necessary to satisfy the target ZMP at the time Tk is obtained as αtmp, assuming that it matches the angular acceleration of the body posture). This αtmp is obtained by using the above equation 03 related to the dynamic model of FIG. 5, for example.

More specifically, for example, the vertical acceleration of the supporting leg mass point 2m and the free leg mass point 2m at the current time Tk is obtained using the time series of the desired foot position / posture obtained up to the current time Tk, and the current From the target foot position / posture at time Tk, the vertical positions of the supporting leg mass point 2m and the free leg mass point 2m are obtained. Further, the vertical position of the body mass point 24m is obtained using the target body vertical position at the current time Tk, and the time series value of the target body vertical position obtained up to the current time Tk is used to obtain the upper position at the current time Tk. The vertical acceleration of the mass point 24m is obtained. Then, these calculated values are substituted into the above equation 03, and My body, d2θby / dt2 in the equation 03 is calculated from an equation obtained by matching “0” and β0, respectively, and the body mass horizontal acceleration d2Xb / dt2 is obtained as the body horizontal acceleration αtmp. In this case, in this embodiment, since the reference body posture angle of the one-leg hopping normal gait is a constant value as described above, the reference body posture angular acceleration β0 is always “0”. The body horizontal acceleration αtmp may be obtained in an exploratory manner so that the horizontal component of the floor reaction force moment around the target ZMP is set to “0” using a dynamic model.

  Next, proceeding to S506, the gait generator 100 obtains the floor reaction force horizontal component Fxtmp at time Tk when the body horizontal acceleration is αtmp using the dynamic model. In the present embodiment, Fxtmp is obtained using Equation 02 of the dynamic model. That is, Fxtmp is obtained by the following equation 17. Here, d2Xsup / dt2 and d2Xswg / dt2 represent the support leg foot mass horizontal acceleration and the free leg foot mass horizontal acceleration at time Tk, respectively.

Fxtmp = mb * αtmp + msup * d2Xsup / dt2 + mswg * d2Xswg / dt2

Next, proceeding to S508, the gait generator 100 determines the magnitude relationship between Fxtmp and the upper limit value Fxmax and the lower limit value Fxmin of the floor reaction force horizontal component allowable range at the current time Tk. Then, according to the determination result, the gait generator 100 determines the body horizontal acceleration α, the floor reaction force horizontal component Fx generated thereby, and the body posture angular acceleration β as follows ( S508 to S516).

That is,
If Fxtmp> Fxmax, the process proceeds to S510, and Fx is determined by the following equation.
Fx = Fxmax ...... Formula 18
Also,
If Fxtmp <Fxmin, the process proceeds to S512, and Fx is determined by the following equation.
Fx = Fxmin ...... Formula 19
If Fxtmp exists in the floor reaction force horizontal component allowable range [Fxmin, Fxmax], the process proceeds to S514, and Fx is determined by the following equation.
Fx = Fxtmp ...... Formula 20

In either case, the process then proceeds to S516, and the body horizontal acceleration α and the body posture angular acceleration β are determined by the following equations 21 and 22.
α = αtmp + (Fx−Fxtmp) / ΔFp (Formula 21)
β = β0 + (αtmp−α) * ΔMp / ΔMr Equation 22

Here, ΔFp is the amount of change in the floor reaction force horizontal component when the body horizontal acceleration is changed by the unit acceleration, and ΔMp is the floor reaction around the target ZMP when the body horizontal acceleration is changed by the unit acceleration. The change amount ΔMr of the horizontal component of the force moment is the horizontal component of the floor reaction force moment around the target ZMP when the body posture angular acceleration is changed by the unit angular acceleration without perturbing the entire center of gravity of the robot 1. The amount of change. In the present embodiment, these ΔFp, ΔMp, and ΔMr are values determined by the following equations 02a, 03a, and 04a based on the equations 02, 03, and 04 related to the dynamic model of FIG.

ΔFp = mb Equation 02a
ΔMp = mb * (Zb−Zzmp) …… Formula 03a
ΔMr = J ...... Formula 04a

Supplementally, if the accuracy of the dynamics calculation is to be increased, the body posture angular acceleration β is calculated as described above, and then the body posture angular acceleration β and the body horizontal acceleration α are combined. The body horizontal acceleration α that can satisfy the target ZMP is preferably determined analytically or exploratively using a stricter dynamic model. In this case, as an exploratory determination method, for example, a pseudo Jacobian (sensitivity matrix) is obtained, and a next candidate is determined by a pseudo Newton method, a simplex method, or the like may be used.

  Further, in order to prevent the floor reaction force horizontal component Fx from strictly exceeding the floor reaction force horizontal component allowable range [Fxmin, Fxmax], in S510, Fx = Fxmax and the floor reaction force moment around the target ZMP. Alternatively, the set of the body horizontal acceleration α and the body angular acceleration β may be obtained in an exploratory manner so that the horizontal component becomes “0”. Similarly, in S512, a set of α and β may be obtained in an exploratory manner so that Fx = Fxmin and the horizontal component of the floor reaction force moment around the target ZMP is “0”. .

  When Fxtmp falls within the allowable range [Fxmin, Fxmax] by the above processing of S504 to S514, Fxtmp becomes Fx as it is. In addition, when Fxtmp exceeds the upper limit value Fxmax of the allowable range [Fxmin, Fxmax] or falls below the lower limit value Fxmin, Fx is forcibly limited to Fxmax and Fxmin, respectively. In particular, in the aerial phase of the one-leg hopping normal gait, Fxmax = Fxmin = 0, so that Fx = 0.

  The body posture angular acceleration β is a floor reaction force that cancels the floor reaction force moment generated around the target ZMP by limiting the body horizontal acceleration α so that Fx does not exceed the allowable range [Fxmin, Fxmax]. It is determined to generate a moment.

  The above is the processing when the time Tk is the time within the period from the time Ts to Tm, or the time within the period from the time Ts2 to Tm2.

  When the determination result in S502 is negative (when Tm ≦ Tk <Ts2 or Tm2 ≦ Tk ≦ Te), the gait generator 100 executes the following processing. First, proceeding to S518, the body posture angular velocity is set to the initial value (at time Ts if current time k is before time Ts2), or by time Ts2 if current time Tk is before time Ts2, otherwise it is before time Ts2. ZMP converted value of the floor reaction force moment that generates the body posture angular acceleration to return to the value of (or otherwise at time Ts2) (hereinafter referred to as the body posture restoring moment ZMP converted value, abbreviated as ZMPrec) ) Pattern.

  In this case, the processing at the time when the current time Tk is after the time Tm2 is the same as the processing at the time before the time Ts2. Therefore, the processing when the current time Tk is before the time Ts2 will be representatively described below.

  The value at time Tk in the body posture angular acceleration pattern for returning the body posture angular velocity to the initial value (value at time Ts) by time Ts2 after time Tm is β (k).

  At this time, floor reaction force moment β (k) * ΔMr is generated due to β (k). As a result, if the floor reaction force vertical component at time Tk is Fz (k), the horizontal position of ZMP (k) calculated from the motion of the robot 1 is the target by ΔZMP (k) obtained by the following equation (23). It will deviate from ZMP.

ΔZMP (k) = − β (k) * ΔMr / Fz (k) Equation 23

Therefore, if the pattern of ΔMr and the pattern of Fz (k) are determined (if known), the body posture angular acceleration pattern that satisfies Expression 23 is set by appropriately setting the pattern of ΔZMP (k). And the body posture angular velocity can be returned to the body posture angular velocity in the initial (time Ts) state of the reference body posture trajectory.

  The body posture restoring moment ZMP converted value (ZMPrec) means ΔZMP (k) appropriately set as described above. Note that ΔMr in Equation 23 is not strictly a constant value, but is approximately a constant value (= J) in the present embodiment.

  In this case, the relationship between β (k) and ZMPrec is given by

β (k) = − ZMPrec * Fz (k) / ΔMr Equation 24

Hereinafter, this β (k) is referred to as an angular acceleration for body posture restoration.

  In the present embodiment, the ZMPrec pattern shape is a trapezoidal shape as shown in FIG. 18, for example. Then, the gait generator 100 sets the integral value of the body posture angular acceleration β from time Ts to Ts2 to “0” in order to make the body posture angular velocity at time Ts2 coincide with the body posture angular velocity at time Ts. The parameters that define the ZMPrec pattern (such as the height of the trapezoid) are determined so that “

  That is, during the period from time Tm to time Ts2, an integral value obtained by integrating the angular acceleration of the sum of β (k) determined according to ZMPrec by Equation 24 and the reference body posture angular acceleration β0 is: ZMPrec so as to coincide with a value obtained by multiplying the integral value of the body posture angular acceleration β (angular acceleration obtained as described above in S504 to S516 in FIG. 17) by −1 in the period from time Ts to time Tm. Parameters defining the pattern are determined.

  In this case, in the present embodiment, the gait generator 100 sets the time of the break point in the trapezoidal pattern of ZMPrec as the prescribed (known) time between the times Tm and Ts2, and the trapezoidal shape. Using the height of the pattern as an unknown parameter that defines the pattern, the height of the ZMPrec pattern that satisfies the above relationship is obtained. As a result, the height of the trapezoidal pattern is calculated as a parameter that defines the ZMPrec pattern.

  Next, in S520, the gait generator 100 calculates the body posture restoring moment ZMP converted value ZMPrec at time Tk based on the ZMPrec pattern determined as described above.

  Next, in S522, the gait generator 100 uses the ZMPrec value at time Tk to calculate the body posture restoration angular acceleration β (k) calculated by the above equation 24 to the reference body posture angle at time Tk. By adding to the acceleration β0, the body posture angular acceleration β at the time Tk is calculated.

  That is, β is calculated by the following equation 24a.

β = β0 + β (k)
= Β0−ZMPrec * Fz (k) / ΔMr …… Formula 24a

Next, in S524, by executing the same processing as S504, it is assumed that the body posture angular acceleration at the time Tk matches the reference body posture angular acceleration β0 at the time Tk, and the target ZMP at the time Tk. The body horizontal acceleration necessary to satisfy the above is obtained as αtmp.

  Next, in S526, the gait generator 100 calculates the horizontal body acceleration α using the following equation 25.

α = αtmp− (ΔMr / ΔMp) * (β−β0)
= Αtmp− (ΔMr / ΔMp) * β (k)
...... Formula 25

In step S528, the gait generator 100 uses the dynamic model to determine the floor reaction force horizontal component Fx at time Tk when the horizontal acceleration of the body is α as described above. In this case, Fx is calculated by the following equation 26 obtained by replacing αtmp on the right side of the equation 17 with α.

Fx = mb * α + msup * d2Xsup / dt2 + mswg * d2Xswg / dt2 Equation 26

Supplementally, if the accuracy of the dynamics calculation is to be increased, the body posture angular acceleration β is obtained as described above, and then the body posture angular acceleration β and the body horizontal acceleration α are synthesized in S526. The body horizontal acceleration α that can satisfy the target ZMP by movement is preferably determined analytically or exploratively using a stricter dynamic model.

  The processing of S518 to S528 described above is processing when the determination result of S502 is negative (when Tm ≦ Tk <Ts2 or Tm2 ≦ Tk ≦ Te).

  In this case, since the floor reaction force horizontal component allowable range is sufficiently large in the period from time Tm to Ts2 and in the period from time Tm2 to Te, the body posture angular velocity is returned while satisfying the target ZMP. Even if the body posture angular acceleration β is determined, it is possible to avoid the floor reaction force horizontal component Fx from deviating from the allowable range [Fxmin, Fxmax].

  Next, the process proceeds to S414 in FIG. 16, and the gait generator 100 obtains the body horizontal velocity at time Tk by sequentially integrating the body horizontal acceleration obtained in S412 (accumulating from time Ts to the current time Tk). . Further, the gait generator 100 obtains the body horizontal position at the time Tk by sequentially integrating the body horizontal speed (accumulating from the time Ts to the current time Tk). The gait generator 100 obtains the body posture angular velocity at time Tk by sequentially integrating the body posture angular acceleration obtained in S412 (accumulating from time Ts to the current time Tk). Further, the gait generator 100 obtains the body posture angle at the time Tk by sequentially integrating the body posture angular velocity (accumulating from the time Ts to the current time Tk).

  After executing the process of S306 in FIG. 15 for determining the instantaneous value of the normal gait as described above, the process proceeds to S308, and the value of the gait generation time Tk is increased by the increment time ΔTk.

  Next, returning to S304, as long as the condition shown in S304 is satisfied, the gait generator 100 repeats the processes of S306 and S308. Then, when the condition shown in S304 is not satisfied, that is, when the generation of the normal gait is completed until the end time Te (= Ts + Tcyc), the process proceeds to S310.

  In S310, the gait generator 100 determines the initial (time Ts) body posture angle and the initial (time Ts) body posture angular velocity from the normal gait generated by the processing of S300 to S308 as described above (hereinafter, referred to as “posture gait”). This is corrected based on the body posture angle at the end time Te of the temporary gait). This modification is intended to satisfy the boundary condition of the normal gait related to the body posture angle and its angular velocity (the condition that the body posture angle and angular velocity are the same at the beginning and end of the normal gait). is there.

  In the present embodiment, this correction is performed according to the following equations 30 and 31, for example.

New initial body posture angular velocity = initial body posture angular velocity of provisional gait− (terminal body posture angle of provisional gait−initial body posture angle of provisional gait) / Tcyc
...... Formula 30

New initial body posture angle = reference body posture angle initial value ...... Equation 31

“Initial” and “end” in these expressions 30 and 31 mean time Ts and Te, respectively.

  In this embodiment, the initial body posture angle is set to be equal to the reference body posture angle initial value as shown in Equation 31, but for example, the average value of the body posture angle of the normal gait is the reference value. The initial body posture angle may be set so as to match the body posture angle initial value. Alternatively, the initial body posture angle may be set such that the average value of the maximum and minimum body posture angles of the normal gait matches the reference body posture angle initial value.

  After completing the process of S310 of FIG. 15, the process proceeds to S210 of FIG. In S210, the gait generator 100 determines the final body horizontal position velocity (body horizontal position and body horizontal velocity at time Te) of the normal gait (provisional gait) generated in S208 at the moment. Is converted into a value viewed from the support leg coordinate system corresponding to the support leg, and the value is (Xe, Vxe) (Xe: end body horizontal position, Vxe: end body horizontal speed).

  Next, in S212, the gait generator 100 calculates the difference between the initial body horizontal position speed (Xs, Vxs) and the terminal body horizontal position speed (Xe, Vxe) as the boundary condition error (errx, errv). . In the normal gait, the boundary condition needs to be satisfied, so the boundary condition error (errx, errv) must be “0” or almost “0”.

  Next, in S214, the gait generator 100 determines whether or not the boundary condition error (errx, errv) calculated in S212 is within a preset allowable range. Instead of determining whether or not the boundary condition error (errx, errv) is within the allowable range in this way, the divergence component defined by the expression 10 and the “+” on the right side of the expression 10 are expressed as “−. And the difference between the initial divergence component (= Xs + Vxs / ω0) and the terminal divergence component (= Xe + Vxe / ω0), and the initial convergence component (= Xs−Vxs / ω0). It may be determined whether the difference between the terminal convergence component (= Xe−Vxe / ω0) is within a certain allowable range.

  If the determination result in S214 is negative, the process proceeds to S216 in order to search for (Xs, Vxs) where the boundary condition error (errx, errv) is “0” or almost “0” (Xs, Vxs). In S216, the gait generator 100 determines a plurality (two in this embodiment) of initial value candidates (Xs + ΔXs, Vxs) and (Xs, Vxs + ΔVxs) in the vicinity of (Xs, Vxs). Here, ΔXs and ΔVxs mean predetermined minute change amounts with respect to Xs and Vxs, respectively. Then, with each of these initial value candidates as the initial state of the body horizontal position speed, a normal gait is generated by the same processing as in S208. Further, values (Xe + ΔXe1, Vxe + ΔVxe1), (Xe + ΔXe2,) obtained by converting the terminal body position speed of the generated normal gait into values viewed from the support leg coordinate system corresponding to the support leg at the moment (time Te). Vxe + ΔVxe2) is obtained. Here, (Xe + ΔXe1, Vxe + ΔVxe1) means the terminal body position velocity corresponding to (Xs + ΔXs, Vxs), and (Xe + ΔXe2, Vxe + ΔVxe2) means the terminal body position velocity corresponding to (Xs, Vxs + ΔVxs). . In the normal gait (provisional gait) generation process in this case, the initial state of the variable other than the body horizontal position speed (state at time Ts) is, for example, an initial value candidate of the body horizontal position speed ( Xs, Vxs). Further, in S216, by the same process as in S210, the difference between each initial value candidate and the corresponding terminal body position velocity, that is, each of the initial value candidates (Xs + ΔXs, Vxs) and (Xs, Vxs + ΔVxs) is determined. Corresponding boundary condition errors (errx, errv) are determined.

  Subsequently, the gait generator 100 proceeds to S218, on the basis of the body horizontal position velocity boundary condition error for each of (Xs, Vxs) and initial value candidates (Xs + ΔXs, Vxs) and (Xs, Vxs + ΔVxs) in the vicinity thereof. The next initial value candidate of (Xs, Vxs) is determined by a search method (a method of obtaining a pseudo Jacobian (sensitivity matrix) and determining the next candidate by the steepest descent method or the simplex method).

  That is, the body horizontal position and the body horizontal velocity are respectively set to the initial value candidates (Xs, Vxs) and the initial value candidates (Xs, Vxs) and the initial value candidates (Xs, Vxs, Vxs) and (Xs, Vxs + ΔVxs). A sensitivity matrix indicating the degree of change in the body horizontal position / velocity boundary condition error when it is slightly changed from Vxs) is obtained, and based on the sensitivity matrix, an initial value candidate (Xs) that makes the boundary condition error smaller , Vxs) is newly determined. Then, after the new initial value candidate (Xs, Vxs) of the body horizontal position speed is determined in this way, the process returns to S206.

  As long as the determination result in S214 is negative, the above processing (the processing in S206 to S218) is repeated. In this case, in S300 in the normal gait generation process (S208) corresponding to the new initial value candidate (Xs, Vxs) of the body horizontal position speed, the initial value of the body posture angular velocity is the reference body posture. Instead of the initial value of the angular velocity, it is set to the value determined in S310 in the processing of S208 corresponding to the previous initial value candidate (Xs, Vxs) of the body horizontal position velocity. If the determination result in S214 is affirmative, the process exits the repeat loop (S204) and proceeds to S220. Note that the provisional gait generated immediately before exiting the repeat loop of S204 is obtained as a normal gait that satisfies the boundary condition.

  In S220, the gait generator 100 determines the initial body horizontal position velocity (X0, Vx0) at the original initial time “0” (the end time of the current gait) and the initial body vertical position velocity at the initial time “0”. (Z0, Vz0) and the initial body posture angle and angular velocity at the initial time “0” are obtained.

  Specifically, (X0, Vx0) and (Z0, Vz0) are the time instants when switching from the second turning gait to the first turning gait in S408 and S414 shown in FIG. The body horizontal position velocity and the body vertical position velocity determined when Tk = Tcyc (= Te−Ts), respectively, are the support leg coordinate systems corresponding to the support legs of the first turning gait starting at time Tcyc. It is determined to be a value obtained by converting to a value seen from the above. Similarly, the initial body posture angle and angular velocity at the initial time “0” correspond to the body posture angle and angular acceleration determined at time Tcyc, corresponding to the support leg of the first turning gait from time Tcyc. It is determined to be a value obtained by converting into a value seen from the support leg coordinate system.

  Next, in S222, the gait generator 100 calculates the normal gait initial divergent component q [0] by the following equation 40 corresponding to the equation 10.

q [0] = X0 + V0 / ω0 Equation 40

  Further, proceeding to S224, the gait generator 100 converts the normal gait initial divergent component q [0] into a value viewed from the current time's gait supporting leg coordinate system as q ″ [0]. In addition, the gait generator 100 converts the initial body vertical position velocity (Z0, Vz0) into a value as viewed from the current time's gait support leg coordinate system (Z0 ″, Vz0 ″). Asking.

  Supplementally, (Z0 ″, Vz0 ″) is the body vertical position velocity at the end of the second turning gait as seen from the supporting leg coordinate system of the second turning gait (next gait supporting leg coordinate system of FIG. 7). It matches. Further, q ″ [0] matches the terminal divergence component of the second turning gait as seen from the supporting leg coordinate system of the second turning gait. Therefore, using these matching relationships, (Z0 ″ , Vz0 ″) and q ″ [0] may be calculated.

  Thus, the processing of S024 in FIG. 6, that is, the subroutine processing for determining the initial state of the normal gait is completed.

  Next, the process proceeds to S026 in FIG. 6, and the gait generator 100 determines the gait parameters of the current time's gait (partially determined). The process of S026 is executed according to the flowchart shown in FIG.

  First, in S600, the gait generator 100 determines the foot gait parameters of the current time gait so that the foot position / posture trajectory of the current time gait is continuously connected to the foot position / posture trajectory of the normal gait. . When the current time's gait is a one-leg hopping gait, the foot trajectory parameters are determined as follows.

  The current free gait foot position / posture of the current time's gait is set to the current free leg foot position / posture (previous gait end free leg foot position / posture) viewed from the current time's gait supporting leg coordinate system. Similarly, the current support gait initial support leg foot position / posture is set to the current support leg foot position / posture (previous gait end support leg foot position / posture) viewed from the current time's gait support leg coordinate system.

  Further, the free leg foot position / posture at the end of the current time gait is set to the position / posture of the free leg foot position / posture at the end of the one-leg hopping gait as viewed from the current time's gait supporting leg coordinate system.

  Further, the foot position / posture at the end of the current time's gait is determined based on the basic landing position / posture corresponding to the next time's gait supporting leg coordinate system (the position of the origin of the next time's gait supporting leg coordinates and the directions of the X and Y axes). The specified basic landing position / posture) is set to the position / posture viewed from the current time's gait supporting leg coordinate system.

  Next, proceeding to S602, the gait generator 100 causes the current gait reference body posture trajectory to be continuously connected to the standard gait reference body posture trajectory parameter of the normal gait. Determine trajectory parameters.

  In the present embodiment, when the current time gait and the next time gait are single leg hopping gaits, the reference body posture trajectory parameter of the current time gait is the normal body posture trajectory of the current time gait. It is determined so as to maintain the same constant value as the reference body posture angle in the case. That is, the reference body posture angle around the X axis and the reference body posture angle around the Y axis are set to the above-mentioned one-leg hopping roll direction body posture angle, “0”, and these values are the current time's gait. As a reference body posture trajectory parameter.

  Next, in step S604, the gait generator 100 determines the arm posture trajectory parameters of the current time gait in the same manner as the arm posture trajectory parameters of the normal gait. However, the arm posture trajectory parameter of the current time gait is determined so that the arm posture trajectory of the current time gait is continuously connected to the arm posture trajectory of the normal gait. Note that the arm posture trajectory parameters determined here are the same as when determining the arm posture trajectory parameters of the normal gait, except for those related to the angular momentum change of both arms around the Z axis (or upper body trunk axis). Motion parameters (specifically, parameters defining the trajectory of the relative center of gravity position of both arms with respect to the upper body 24).

  Next, proceeding to S606, the gait generator 100 determines a floor reaction force vertical component trajectory parameter of the current time's gait. If the current time's gait and the next time's gait are single leg hopping gaits, the floor reaction force vertical component trajectory parameters of the current time's gait are determined by the floor reaction force vertical component trajectory defined by the gait as described above. It is determined to have a pattern similar to the floor reaction force vertical component trajectory of each turning gait. That is, the floor reaction force vertical component trajectory parameter of the current time's gait is determined so that the floor reaction force vertical component trajectory changes in the pattern shown in FIG.

  In this case, the floor reaction force vertical component trajectory parameter is determined so that the overall center-of-gravity vertical position velocity of the current time's gait and the floor reaction force vertical component trajectory are continuously connected to the one-leg hopping normal gait.

  Specifically, first, the kinematics model (for example, FIG. 5) of the robot 1 is determined from the initial body position and posture of the normal gait finally obtained by the processing of S024 in FIG. Model) to obtain the initial overall center-of-gravity vertical position velocity of the normal gait viewed from the gait supporting leg coordinate system.

  For example, the initial overall center-of-gravity vertical position of the normal gait viewed from the current time's gait supporting leg coordinate system corresponds to the initial body vertical position of the normal gait viewed from the current time's gait supporting leg coordinate system. From the upper body mass point vertical position and the supporting leg mass point vertical position and the free leg mass point vertical position of the model of FIG. 5 corresponding to each initial foot position of the normal gait viewed from the current time gait supporting leg coordinate system, It is calculated | required by Formula 04.

  Further, the initial overall center-of-gravity vertical velocity of the normal gait viewed from the current time's gait supporting leg coordinate system is based on the model of FIG. 5 corresponding to the upper body vertical velocity of the normal gait viewed from the current time's gait supporting leg coordinate system. From the body point vertical velocity and the supporting leg mass point vertical velocity and free leg mass point vertical velocity of the model of FIG. 5 corresponding to the initial foot vertical velocity of the normal gait as seen from the gait supporting leg coordinate system this time, It is obtained by an expression obtained by differentiating both sides of Expression 04.

  These initial overall center-of-gravity vertical position velocities may be calculated using a more rigorous model from the initial motion state of the normal gait.

  Then, the initial total center-of-gravity vertical position speed of the normal gait obtained in this way is substituted into the terminal total center-of-gravity vertical position speed of Equation 13 and Formula 41 below, and Specifically, the total center of gravity vertical position and the total center of gravity vertical velocity at the end of the previous gait are values substituted for the initial total center of gravity vertical position velocity of Equation 13 and Equation 41 below. The floor reaction force vertical component trajectory parameter of the current time's gait is determined so as to satisfy the relationship of 13 and Equation 41. Note that the integrals in Equation 13 and Equation 41 are integrals during the period from the beginning to the end of the current time's gait.

Final total center-of-gravity vertical velocity-initial total center-of-gravity vertical velocity = first floor integral of (floor reaction force vertical component / robot mass) + first floor integral of gravitational acceleration
... Formula 41

However, the gravitational acceleration of Expression 41 is a negative value.

  More specifically, first, at least two parameters of floor reaction force vertical component trajectory parameters (such as break point time) defining the floor reaction force vertical component pattern as shown in FIG. 20 are set as independent unknown variables. The value of the unknown variable is determined by simultaneous equations consisting of Equation 13 and Equation 41.

  For example, the height of the trapezoid (the peak value of the floor reaction force vertical component) and the width (the time of the one-leg support period) of FIG. In this case, the slopes of both sides of the trapezoid in FIG. 20 are set to predetermined values according to the current time's gait cycle or the like, or the floor reaction force vertical component pattern except the time of transition from the one-leg support period to the aerial period is broken. The point time is set to a predetermined value according to the current time's gait cycle or the like.

  Next, in S608, the gait generator 100 determines that the floor reaction force horizontal component permissible range [Fxmin, Fxmax] (specifically, parameters that define the pattern of the floor reaction force horizontal component permissible range) is a normal gait. Determined in the same manner as the first turning gait or the second turning gait. For example, the floor reaction force horizontal component allowable range is set in a pattern as shown in FIG. In the present embodiment, the floor reaction force horizontal component allowable range is determined based on the above formulas 12a and 12b from the floor reaction force vertical component trajectory previously determined in S606.

  Next, in S610, the gait generator 100 tentatively determines ZMP trajectory parameters that define the ZMP trajectory of the current time's gait.

  In this case, the ZMP trajectory parameter satisfies the requirement that the target ZMP defined thereby has a high stability margin and does not change abruptly, and from the target ZMP at the end of the previous gait, the initial normal gait. It is determined so as to continuously change up to the target ZMP.

  When the current time gait and the next time gait are one-leg hopping gaits, this target ZMP is, for example, the support leg foot 22 (22R in the illustrated example) corresponding to the current time gait support leg coordinate system of FIG. It is determined as shown by the target ZMP trajectory with the support leg foot 22 (22R in the illustrated example) corresponding to the support leg coordinate system.

  In this example, the previous gait is a walking gait (or a running gait), and in the one-leg support period of the current time's gait, the target ZMP is a position closer to the heel of the support leg foot 22 (22R) (previous time). It is set so as to continuously move from the position of the target ZMP at the end of the gait) toward the right front side of the ground contact surface of the supporting leg foot 22. In the aerial period, the target ZMP is determined from the position on the right front side of the ground contact surface of the support leg foot 22 to approximately the center in the ground contact surface of the support leg foot 22 (22R) corresponding to the next time gait support leg coordinate system. It is set so as to continuously move to the point (position of the target ZMP in the initial stage of the normal gait). In this case, the target ZMP in the one-leg support period may be kept constant during a part of the one-leg support period.

  In step S610, the position of the target ZMP immediately before the end of the one-leg support period is determined as the ZMP trajectory parameter of the current time gait.

  When the previous gait is a one-leg hopping gait, the target ZMP may be set in a pattern similar to the first turning gait of the normal gait.

  Supplementally, the ZMP trajectory parameter of the current time's gait determined in S610 is only provisionally determined and is corrected as described later. Therefore, the ZMP trajectory of the current time gait defined by the ZMP trajectory parameters provisionally determined in S610 as described above is hereinafter referred to as a temporary target ZMP trajectory of the current time gait.

  Next, in S612, the gait generator 100 determines the body posture restoration period [Ta, Tb]. The start time Ta of the body posture angle restoration period is equivalent to Tm in the second turning gait of the normal gait, and the end time Tb of the body posture restoration period is Ts2 in the second turning gait of the normal gait. It is equivalent. The way of setting these times Ta and Tb is the same as the way of setting Tm and Ts2. Therefore, the body posture restoration period [Ta, Tb] is determined as illustrated in FIG.

  The above is the process executed in S026.

  Returning to the description of FIG. 6, after performing the process shown in S026 (the process of determining or provisionally determining the gait parameters of the current time's gait) as described above, the process then proceeds to S028, where the gait generating device 100 determines Correct the gait parameter of the gait (confirm the gait parameter this time). The gait parameter to be corrected here is a ZMP trajectory parameter that defines the temporary target ZMP trajectory. In the processing of S028, the ZMP trajectory parameters are corrected so that the body position / posture trajectory is made continuous or asymptotic to the normal gait.

  Specifically, the process of S028 is executed according to the flowchart of FIG.

  First, the process proceeds to S702 via S700, and based on the provisional current time gait parameter, a provisional current time gait up to the end time of the current time gait is generated. The temporary current time gait parameter is a parameter other than the ZMP trajectory parameter that defines the temporary target ZMP trajectory (the temporary target ZMP trajectory determined in S026 or S716 described later) and the ZMP trajectory parameter determined as described above in S026. It means a set with gait parameters.

  Specifically, the process of S702 is executed according to the flowchart of FIG.

  As will be described below, the gait generator 100 performs various initializations in S800. Specifically, “0” is substituted into the provisional gait generation time (temporary current time gait generation time) Tk. Also, the last state of the previous gait (more specifically, the body horizontal position speed, the body vertical position speed, the body posture angle and its angular speed at the end time of the previous gait, the desired foot position / posture, the target arm) The motion state such as the posture) converted to the state viewed from the current support leg coordinate system is set as the initial state of the provisional current time gait.

  Next, in step S804 through step S802, the gait generator 100 determines whether the provisional gait generation time Tk is a time before the end time Tcurr of the current time gait (whether Tk ≦ Tcurr). To do. If the determination result is affirmative, the process proceeds to S806, and the gait generator 100 determines the instantaneous value (value at time Tk) of the current time gait. Next, in S808, the gait generator 100 increases the temporary gait generation time Tk by a predetermined step time ΔTk, and then makes the determination in S804 again. Note that the step time ΔTk is, for example, a time that coincides with the control cycle Δt.

  If the determination result in S804 is negative, the process of the flowchart shown in FIG. 23 is completed.

  Through the above processing, a provisional current time gait is generated from the initial stage to the end.

  Specifically, the process of S806 is executed according to the flowchart of FIG.

  First, in S900, the gait generator 100 is based on the gait parameters (specifically, the floor reaction force vertical component trajectory parameters of the provisional current time gait) at the time Tk of the desired floor reaction force vertical component shown in FIG. Find the instantaneous value.

  Next, in S902, the gait generator 100 obtains an instantaneous value of the temporary target ZMP at time Tk based on the gait parameters (specifically, the ZMP trajectory parameters of the temporary current time gait).

  Next, in step S904, the gait generator 100 determines the desired foot position / posture and reference at time Tk based on the gait parameters (specifically, foot trajectory parameters, reference body posture trajectory parameters, and arm posture trajectory parameters). Instantaneous values of upper body posture and target arm posture are obtained. However, regarding the target arm posture, the position of the center of gravity of both arms is determined in more detail, but the movement of the arm (arm swing) that changes the angular momentum around the vertical axis (or the trunk axis of the upper body 24). Exercise) has not yet been determined.

  Note that the instantaneous value of the desired foot position / posture at time Tk is obtained using a finite time settling filter in the same manner as when the foot position / posture at initial time Ts is obtained in S200 of FIG.

  Next, in S906, the gait generator 100 satisfies the desired floor reaction force vertical component obtained in S900 (the sum of the vertical inertia force and gravity of the robot 1 is balanced with the desired floor reaction force vertical component). The overall center-of-gravity vertical position speed is calculated. Specifically, this calculation is performed by the same processing as S406 in FIG.

  Next, in S908, the gait generator 100 calculates an instantaneous value at time Tk of the body vertical position that satisfies the overall center-of-gravity vertical position obtained as described above. Specifically, this calculation is performed by the same process as in S408 of FIG.

  Next, proceeding to S910, the gait generating device 100 determines the floor reaction force horizontal component permissible shown in FIG. 21 based on the gait parameters (specifically, parameters that define the floor reaction force horizontal component permissible range of the provisional current time gait). The instantaneous value at time Tk in the range [Fxmin, Fxmax] is calculated.

  Next, in S912, the gait generator 100 determines the body horizontal acceleration and the body posture angular acceleration at the time Tk of the provisional current time gait. In this case, the body horizontal acceleration and the body posture angular velocity satisfy the target ZMP (specifically, the resultant force of the inertial force generated by the target motion of the robot 1 and the gravity acting on the robot 1 is about the target ZMP. In addition to satisfying the condition that the horizontal component of the moment generated in the robot is “0”), the floor reaction force horizontal component Fx that balances the horizontal inertial force of the robot 1 deviates from the allowable range [Fxmin, Fxmax]. It is determined so as to satisfy the condition of not.

  This process is slightly different from the process of S412 in FIG.

  Specifically, the process of S912 is executed as shown in the flowchart of FIG.

  First, in S1000, the gait generator 100 determines whether or not the time Tk is within the body posture restoration period [Ta, Tb] set in S612 of FIG.

  When this determination result is negative (when the floor reaction force horizontal component permissible range at time Tk is relatively narrow or “0”), the gait generator 100 performs the processing from S1002 to S1014. Execute the process. The processing from S1002 to S1014 is the same as the processing from S504 to S516 in FIG.

  On the other hand, when the determination result in S1000 is affirmative (when the allowable range of the floor reaction force horizontal component at time Tk is relatively wide), the process proceeds to S1016, and the gait generator 100 causes the body posture restoring moment. The pattern of the ZMP conversion value ZMPrec is set. In this case, unlike S518 in FIG. 17 relating to generation of a normal gait, ZMPrec is set to a pattern having a constant value “0” (that is, a pattern in which the height of the trapezoidal pattern shown in FIG. 18 is “0”). The

  Next, the processing from S1018 to S1026 is performed. The processing from S1018 to S1026 is the same as the processing from S520 to S528 in FIG.

  In the processing of S1016 to S1026, since the instantaneous value of the body posture restoring moment ZMP converted value ZMPrec is always “0”, the body posture angular acceleration β obtained in S1020 is the reference body posture at time Tk. It corresponds to the angular acceleration β0. Therefore, it is possible to set only β = β0 and perform only the processing of S1022, S1024, and S1026. Further, since the body horizontal acceleration α calculated in S1024 always coincides with αtemp, the processing in S1024 may be omitted.

  After completing the process of S912 in FIG. 24 as described above, the process then proceeds to S914, where the gait generator 100 sequentially integrates the body horizontal acceleration obtained in S912 (from the initial time “0” of the current time gait to the current time). The body horizontal velocity at time Tk is obtained by integrating until time Tk). Furthermore, the gait generator 100 obtains the body horizontal position at the time Tk by sequentially integrating the body horizontal velocity (accumulating from the initial time “0” of the current time gait to the current time Tk). Further, the gait generator 100 successively integrates the body posture angular acceleration obtained in S912 (accumulates from the initial time “0” of the current time gait to the current time Tk) to thereby obtain the body posture angular velocity at the time Tk. Ask for. Further, the gait generator 100 obtains the body posture angle at the time Tk by successively integrating the body posture angular velocity (accumulating from the initial time “0” of the current time gait to the current time Tk). .

  As described above, after the processing of S702 in FIG. 22 (provisional current time gait generation processing up to the end time) is completed, the process proceeds to S704. In S704, the gait generator 100 determines the body horizontal position speed (Xe, Vxe) at the end of the provisional current time gait obtained in S702 (Xe: body horizontal position at the end of the provisional current time gait, Vxe: provisional. The terminal divergent component qe1 (= Xe + Vxe / ω0) of the provisional current time gait is calculated from the current body gait end body horizontal velocity) on the basis of the equation (10).

  Next, the processing proceeds to S706, where the gait generator 100 determines the terminal divergence component error which is the difference between the terminal divergence component qe1 of the provisional current time gait and the initial divergence component q0 "of the normal gait previously obtained in S224 of FIG. errq (= qe1−q0 ″) is calculated.

  Next, proceeding to S708, the gait generator 100 determines whether or not the calculated terminal divergence component error errq is within an allowable range (a predetermined range near “0”).

  If the determination result in S708 is negative, the process proceeds to S710, and the gait generator 100 uses the ZMP correction amount shown in FIG. 26 set as a = Δa (Δa is a predetermined minute amount) as the current temporary value. Using the target ZMP trajectory obtained by correcting the target ZMP trajectory, a temporary current time gait up to the end time is generated as in S702.

  Here, referring to FIG. 26, the ZMP correction amount matches the terminal divergence component of the current time gait with the initial divergence component of the normal gait as much as possible (as a result, the body position / posture trajectory of the current time gait in the future This means a correction amount for correcting (adding to the temporary target ZMP) the temporary target ZMP in order to converge to the body position / posture trajectory of the normal gait. In this embodiment, the ZMP correction amount is a trapezoid pattern correction amount, and the height of the trapezoid pattern is “a”.

  In this case, in the present embodiment, the correction of the temporary target ZMP by the ZMP correction amount of the trapezoid pattern is the body posture restoration period [Ta, Tb] as a period in which the floor reaction force horizontal component allowable range is sufficiently wide. The trapezoid pattern is set within the body posture restoration period [Ta, Tb]. Note that a = Δa is set in step S710 in order to observe a change in the terminal divergence component error errq when the current temporary target ZMP trajectory is corrected by a minute amount by the ZMP correction amount of the trapezoid pattern.

  Note that Δa may be a small amount of constant, but Δa may be set so as to decrease Δa as the terminal divergence component error errq is reduced by repeated calculation described below.

  As described above, after generating the temporary current time gait using the temporary target ZMP trajectory corrected by the ZMP correction amount in S710, the process proceeds to S712. In S712, the gait generator 100 determines the horizontal horizontal position velocity (Xe1, Vxe1) at the end of the provisional current time gait generated in S710 (Xe1: the final body horizontal position of the provisional current time gait, Vxe1: provisional. Based on Equation 10 above, the terminal divergent component qe2 (= Xe1 + Vxe1 / ω0) of the provisional current time gait (the provisional current time gait generated in S710) is calculated from the current body gait end body horizontal velocity).

  Next, proceeding to S714, the gait generator 100 calculates the parameter sensitivity r (the rate of change of the terminal divergence component error errq with respect to Δa) using the formula shown in the figure. That is, the parameter sensitivity r is calculated by dividing the difference between qe2 and qe1 by Δa.

  Next, the process further proceeds to S716, where the gait generator 100 increases the value obtained by dividing the terminal divergence component error errq obtained in S706 by the parameter sensitivity r obtained in S714, a = −errq / r. By setting the ZMP correction amount of the trapezoid pattern by setting a, the temporary target ZMP trajectory (the temporary target ZMP trajectory used for generating the temporary current time gait in the processing of S702) is corrected by this ZMP correction amount. A new temporary target ZMP trajectory is determined.

  Next, the gait generator 100 executes the processing from S702 again. In this case, in S702, a temporary current time gait is generated using the new temporary target ZMP trajectory determined in S716.

  Then, as long as the determination result in S708 is negative, the processing in S702 to S716 described above is repeated. If the determination result in S708 is positive, the process goes through the repetition loop and proceeds to S718.

  By repeating the processes of S702 to S716, the temporary target ZMP trajectory is corrected so that the terminal divergent component of the temporary current time gait matches or substantially matches the initial divergent component of the normal gait.

  Next, in S718, the gait generating device 100 determines the current time gait based on the temporary current time gait (the temporary current time gait generated in S702 immediately before exiting the repeat loop of S702 to S716) and the normal gait. The pattern of the body posture restoring moment ZMP converted value (ZMPrec) of the current time's gait is determined so that the body posture angle of the current time gait approaches the body posture angle of the normal gait.

  In this case, the pattern of the body posture restoring moment ZMP converted value (ZMPrec) is 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 of the provisional current time gait. It is determined based on the difference between the body posture angular velocity and the initial body posture angular velocity of the normal gait.

  The ZMPrec determined here is a period during which the floor reaction force horizontal component allowable range is sufficiently large in the instantaneous value generation process of the current time gait described later (the process of S032) (the body posture restoration period [Ta, Tb]), even if the body posture angular acceleration is generated so that the body posture angular trajectory can be connected to the normal gait, the consistency between the current gait end divergence component and the normal gait initial divergence component This is for correcting the temporary target ZMP so that (condition of S708) can be maintained.

  This ZMPrec is a trapezoidal pattern similar to that described in the normal gait generation process, and is determined as follows. That is, in the trapezoidal pattern of ZMPrec illustrated in FIG. 18, the time of the apex (folding point) of the trapezoid is a predetermined time within the body posture restoration period [Ta, Tb] (for example, the trapezoidal pattern of the ZMP correction amount). (Time of break point). Then, using the trapezoidal height as an unknown parameter, the trapezoidal height of ZMPrec is obtained as follows. In S718, Tm and Ts2 in FIG. 18 are replaced with Ta and Tb, respectively.

  As described above, when there is one unknown parameter of the pattern of the body posture restoring moment ZMP converted value ZMPrec, both the body posture angle and the body posture angular velocity at the end of the current time's gait are steady gaits. In general, it is not possible to connect them continuously. Therefore, in the present embodiment, the unknown parameter is determined so that the gait state generated gradually approaches the normal gait state in a plurality of gait cycles.

  Supplementally, by making the ZMPrec pattern more complex, the number of unknown parameters of the pattern may be two or more. Then, two or more unknown parameters may be determined so that both the body posture angle and the body posture angular velocity are continuously connected to the normal gait at the end of the current time's gait. However, in this case, there is a possibility that the ZMPrec pattern fluctuates too much in a zigzag manner.

  The trapezoidal height as an unknown parameter of ZMPrec is determined as described below. Hereinafter, after describing the calculation principle, the calculation procedure will be described.

  As described above, the difference between the terminal body posture angle of the provisional current time gait generated with ZMPrec = 0 in S702 and the initial body posture angle of the normal gait is defined as θerr. 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 defined as Vθerr.

  Here, it is assumed that the current time gait is generated with the trapezoidal height of the ZMPrec pattern as a certain value bcurr, and then the first turning gait of the normal gait is generated by the same algorithm as the current time gait. However, the height of the body posture restoring moment ZMP converted value ZMPrec pattern of the first turning gait is the height of the ZMPrec pattern of the first turning gait obtained in S518 of FIG. 17 (the trapezoid pattern illustrated in FIG. 18). ) And a certain value b1.

  The gait thus generated is called a ZMPrec corrected gait, and the body posture angle and its angular velocity at its end (the end of the first turning gait) are θ1 and Vθ1, respectively.

  The original normal gait (finally the initial body posture angle of the normal gait determined in S310 and the initial body posture angle determined at the time when the process of S024 in FIG. 6 (process for determining the initial state of the normal gait) is completed. The normal body posture angle of the first turning gait of the normal turning gait when the angular velocity is the initial value and the ZMPrec patterns of the first turning gait and the second turning gait are the patterns obtained in S518. And angular velocity are θ1org and Vθ1org, respectively.

  Here, Δθ1 and ΔVθ1 are defined as follows.

Δθ1 ＝ θ1−θ1org …… Formula 50
ΔVθ1 = Vθ1−Vθ1org ...... Formula 51

Δθ1 means the difference between the body posture angle at the end of the ZMPrec corrected gait and the body posture angle at the end of the first turning gait of the original normal gait. Similarly, Vθ1 means the difference between the body posture angular velocity at the end of the ZMPrec corrected gait and the body posture angular velocity at the end of the first turning gait of the original normal gait. If Δθ1 and ΔVθ1 become “0”, the ZMPrec pattern is matched with the ZMPrec for the second turning gait obtained in S518 by the same algorithm as the current time gait following the ZMPrec corrected gait. If the second turning gait is generated, the body posture angle and angular velocity of the second turning gait coincide with those of the second turning gait of the original normal gait.

  Accordingly, bcurr (hereinafter referred to as the current time gait trapezoidal height bcurr) and b1 (hereinafter referred to as the first turning gait trapezoidal height b1) at which Δθ1 and ΔVθ1 are “0” are obtained, and the obtained bcurr is determined as the current step. The final trapezoidal height may be determined.

  On the other hand, since the dynamic model related to the body posture angle of the robot 1 has linear characteristics like the flywheel FH shown in FIG. 5, Δθ1 and ΔVθ1 are the gait trapezoid height bcurr this time, the first turning gait. Trapezoidal height b1, the difference between the terminal body posture angle of the temporary gait and the initial body posture angle of the normal gait θerr, the terminal body posture angular velocity of the temporary gait and the initial body posture angular velocity of the normal gait And the following relationship with the difference Vθerr.

Δθ1 = c11 * bcurr + c12 * b1 + θerr + e1 * Vθerr …… Formula 52
ΔVθ1 = c21 * bcurr + c22 * b1 + e2 * Vθerr …… Formula 53

However, c11, c12, c21, c22, e1, e2 are the gait cycle of the current time's gait, the gait cycle of the first turning gait, and the pattern parameters of the body posture restoring moment ZMP converted value ZMPrec (especially regarding time) The coefficient is uniquely determined by the parameter).

  Based on the above principle, the gait generator 100 determines the trapezoid height as an unknown parameter of ZMPrec as follows.

  First, the gait generator 100 obtains a body posture angle difference θerr and a body posture angular velocity difference Vθerr at the boundary between the provisional current time gait and the normal gait.

  Next, the gait generator 100 uses c11, c12, c21, c22, e1, and e2 as coefficients of the equations 52 and 53 as the gait cycle of the current time gait, the gait cycle of the first turning gait, and the upper body. It is determined based on the pattern parameters (particularly parameters relating to time) of the posture restoring moment ZMP converted value ZMPrec.

  Next, the gait generator 100 determines the current time gait trapezoidal height bcurr and the first turning gait trapezoidal height b1 so that the value of the calculation result on the right side of Equations 52 and 53 is “0”. That is, bcurr and b1 are obtained by solving simultaneous equations with the left side of Expressions 52 and 53 set to “0”.

  The gait generator 100 determines the bcurr obtained as described above as the trapezoidal height of the trapezoid pattern of the body posture restoring moment ZMP converted value (ZMPrec) of the current time's gait.

  As described above, after determining the pattern of the body posture restoring moment ZMP converted value ZMPrec in S718, the process proceeds to S720. In S720, the gait generator 100 sets the body posture restoring moment ZMP converted value pattern determined in S718 to the current temporary target ZMP trajectory (the temporary target ZMP trajectory when the loop of S702 to S716 is repeated). The pattern obtained by superimposing (adding together) is determined as the target ZMP trajectory of the current time's gait.

  As described above, by determining the target ZMP trajectory of the current time gait by superimposing the body posture restoring moment ZMP converted value pattern on the temporary target ZMP trajectory, the horizontal position trajectory of the body is changed to the horizontal position of the temporary current gait. The body posture angular trajectory of the current time's gait will lead to the normal gait, while keeping the trajectory out of the trajectory (and thus keeping the terminal divergence component of the current time's gait almost the same as the initial divergence component of the normal gait) In addition, the gait parameters of the current time's gait are finally determined.

  Returning to FIG. 6, after correcting the current time gait parameter in S028 as described above (after confirming the current time gait parameter), or when the determination result in S016 is negative, the process proceeds to S030 and the gait. The generation device 100 determines an instantaneous value of the current time's gait.

  In S030, the gait generator 100 executes the same processing as S900 to S910 of FIG. 24, and then executes the processing of S1000 to S1026 of FIG. 25 which is the subroutine processing of S912. However, in this case, in S1016 of FIG. 25 executed in the process of S030, as a pattern of the body posture restoring moment ZMP converted value (ZMPrec), S718 of FIG. 22 executed in the process of S028. The pattern of the body posture restoring moment ZMP converted value (ZMPrec) determined in (1) is set.

  Next, the gait generator 100 executes the same processing as S914 in FIG. 24, whereby the processing in S030 in FIG. 6 is completed.

  The time Tk in the process of S030 is the time t of the current control cycle (current control cycle) of the gait generator 100.

  Next, the processing proceeds to S032, and the gait generator 100 cancels the spin force (the floor reaction force moment vertical component (Z-axis component) generated around the target ZMP by a motion other than the arm of the robot 1 is “0” or An arm swinging motion (a motion of swinging the left and right arm bodies in the reverse direction) is determined to make it almost “0”.

  Specifically, the floor reaction force moment vertical component trajectory assuming that the current time's gait was generated so as not to perform arm swinging motion of the arm body (strictly, the current time's gait is generated without shaking the arm body) In this case, the resultant force of the inertial force generated by the motion of the robot 1 and the gravity acting on the robot 1 is obtained by reversing the sign of each instantaneous value of the moment vertical component trajectory acting on the target ZMP). That is, the floor reaction force moment vertical component that balances the instantaneous value of the inertial force moment vertical component generated around the target ZMP by the motion of the current time's gait generated by the processing of S030 (this does not include the arm swing motion). The instantaneous value of is calculated. Then, by dividing the instantaneous value of the vertical component of the floor reaction force moment by the equivalent moment of inertia of the arm swing motion, the angular acceleration of the arm swing motion necessary for canceling the spin force is obtained. If the arm swing is too large, the instantaneous value of the floor reaction force moment vertical component should be divided by a value larger than the equivalent inertia moment of the arm swing motion to obtain the angular acceleration of the arm swing motion. It may be.

  Next, the gait generator 100 performs second-order integration of the angular acceleration of this arm swing motion, and sets the angle obtained by passing this integrated value through the low cut filter as the arm swing motion angle. Thereby, the postures of both arms including the arm swinging motion are determined. However, in the arm swinging operation, the left and right arm bodies are swung in the reverse direction, so that the center of gravity positions of both arm bodies are not changed.

  Note that an arm swing motion for canceling the spin force may be generated even in a normal gait, and the arm swing motion in the current time gait may be determined so as to lead to this.

  Next, proceeding to S034, the gait generator 100 increases the gait generation time t by Δt (control cycle of the gait generator 100), and executes the processing from S014 again.

  The above is the target gait generation process in the gait generator 100.

  Next, operation control of the robot 1 by the control unit 26 will be described with reference to FIG. In the gait generator 100, the target body position / posture trajectory and the target arm posture trajectory among the target gaits generated as described above are output to the robot geometric model (reverse kinematics calculation unit) 102. .

  The target foot position / posture trajectory, target ZMP trajectory (target floor reaction force center point trajectory), and target floor total reaction force trajectory (target floor reaction force horizontal component and target floor reaction force vertical component) The data is output to the determination unit 104 and the target floor reaction force distributor 106.

  Then, the desired floor reaction force distributor 106 distributes the desired total floor reaction force to each foot 22R, 22L, and each desired foot floor reaction, which is the target position of the floor reaction force center point for each foot 22R, 22L. The force center point and each desired foot floor reaction force, which is the desired floor reaction force for each foot 22R, 22L, are determined. The determined desired foot floor reaction force center point and the desired foot floor reaction force are input to the composite compliance operation determining unit 104.

  From the composite compliance operation determination unit 104, the corrected desired foot position / posture trajectory with mechanism deformation compensation is output to the robot geometric model 102. The robot geometric model 102 has 12 joints (10R (L)) of the legs 2 and 2 that satisfy the target body position / posture trajectory and the corrected target foot position / posture trajectory with mechanism deformation compensation. The joint displacement command (the command value of the displacement amount of each joint) is calculated, and this joint displacement command is output to the joint displacement controller 108.

  The joint displacement controller 108 uses the joint displacement command calculated by the robot geometric model 102 as a target value, and controls the displacement amounts of the twelve joints of the robot 1 to follow the target value. Further, the robot geometric model 102 calculates a joint displacement command for each arm body satisfying the target arm posture trajectory and outputs the joint displacement command to the joint displacement controller 108. Then, the displacement controller 108 uses the joint displacement command calculated by the robot geometric model 102 as a target value, and controls the displacement amount of the joint of the arm body of the robot 1 to follow the target value.

  A floor reaction force acting on the robot 1 (specifically, an actual foot reaction force acting on each foot 22) is detected by the control unit 26 based on the output of the six-axis force sensor 36. Is done. The detected value is input to the composite compliance operation determination unit 104.

  The actual body posture angle of the robot 1 (specifically, the actual body posture angle in the roll direction (around the X axis) and the actual body posture angle in the pitch direction (around the Y axis)) outputs the output of the tilt sensor 38. Based on this, the control unit 26 detects the detected value, and the detected value is input to the posture stabilization control calculation unit 112. In this posture stabilization control calculation unit 112, the target total floor reaction force center is used as a floor reaction force operation amount to be additionally applied to the robot 1 in order to restore the actual body posture angle of the robot 1 to the target body posture angle. The compensated total floor reaction force moment around the point (target ZMP) is calculated by a feedback control law such as the PD law. This compensated total floor reaction force moment is input to the composite compliance operation determination unit 104.

  The composite compliance operation determining unit 104 corrects the target floor reaction force based on the input value. Specifically, the target floor reaction force is corrected so that the compensated total floor reaction force moment acts around the target total floor reaction force central point (target ZMP). Then, the composite compliance action determination unit 104 follows the target foot position / posture generated by the gait generator 100 while causing the actual foot position / posture of the robot 1 to follow the target floor reaction force corrected. The corrected desired foot position / posture trajectory with mechanism deformation compensation is determined so as to follow the floor reaction force. However, since it is practically impossible to match all the motion states of the robot 1 and the floor reaction force to the target, a trade-off relationship is given between them to make them coincide as much as possible. That is, the control deviation for each target is given a weight, and control is performed so that the weighted average of the absolute value or the square value of the control deviation is minimized. As a result, the actual foot position / posture and the floor reaction force are approximately equal to the target foot position / posture (the target foot position / posture generated by the gait generator 100) and the corrected target floor reaction force, respectively. It is controlled to follow.

  The configuration and operation of the composite compliance operation determination unit 104 and the like described above are described in detail in Japanese Patent Application Laid-Open No. 10-277969 filed earlier by the applicant of the present application. Keep on.

  Supplementally, in the present embodiment, the joint control means in the present invention is realized by a portion of the function of the control unit 26 shown in FIG. 3 excluding the gait generator 100.

  According to the present embodiment described above, when the robot 1 performs a one-leg hopping operation (when the gait generator 100 generates a one-leg hopping gait), the reference in the roll direction (around the X axis) As the body posture angle, the one-leg hopping roll direction body posture angle shown in FIG. 9 is set. For this reason, the target body posture angle in the roll direction in the landing state of the support legs after the aerial period is equal to or equal to the roll direction body posture angle for one leg hopping as the reference body posture angle. A close attitude angle will be determined. Here, the target body posture angle in the roll direction in the landing state of the support legs after the aerial period does not necessarily match the reference body posture angle. This is because the body posture angle is corrected.

  Therefore, in the landing state of the support leg after the mid-air period of the one-leg hopping operation, the target body posture angle is set to be a free leg rather than the base end portion (hip joint) of the support leg-side leg 2 as shown in FIG. The posture angle is such that the upper body 24 is inclined so that the base end (hip joint) of the side leg 2 is at a relatively high position. For this reason, the target gait (one-leg hopping gait) of the robot 1 is generated so that the entire center of gravity Ga of the robot 1 exists at a position immediately above the hip joint of the supporting leg side leg 2 or a position close thereto. Become.

  In addition, since the target gait is generated so as to satisfy the target ZMP set in the ground contact surface of the support leg foot 22 in the landing state of the support leg, the above-mentioned immediately after landing of the support leg side leg 2 is performed. As a result, the overall center of gravity Ga is located almost immediately above the ground contact surface of the support leg foot 22.

  As a result, by controlling the operation of the robot 1 so as to follow this target gait (single leg hopping gait), the landing leg movement of the supporting leg side leg 2 immediately after the mid-air period of the one leg hopping action is performed. Thus, it is possible to prevent a large moment (moment about the X axis) from acting on the hip joint of the supporting leg-side leg 2 while satisfying the dynamic equilibrium condition of the robot 1. As a result, the burden on the joint actuator of the hip joint (the electric motor 42 in this embodiment) can be reduced, the joint actuator and the hip joint can be reduced in size or weight, and the durability of the joint actuator and the hip joint can be increased. be able to.

  In the embodiment described above, the generation process of the walking gait or the running gait is the same as that described in the embodiment of Patent Document 1, and thus the reference in these walking gait or the running gait. The body posture is, for example, a body posture when the robot 1 is standing in an upright posture (a posture in which the trunk axis is directed in the vertical direction).

  In this case, the reference body posture trajectory is changed at the transition from the one-leg hopping gait to the running gait or the walking gait, or at the transition from the running gait or the walking gait to the one-leg hopping gait. It is desirable to set so that the reference body posture trajectory corresponding to the previous gait continuously changes to the reference body posture trajectory corresponding to the gait after the transition.

  Specifically, for example, when the current time's gait is a one-leg hopping gait and the next time's gait is a walking gait or a running gait, a reference in the roll direction (around the X axis) of the current time's gait. As shown in FIG. 27, the body posture trajectory is preferably set so as to smoothly change from the body posture angle in the roll direction for one leg hopping to “0”.

  Also, for example, when the current time's gait is a walking gait or a running gait and the next time's gait is a one-leg hopping gait, the reference body posture in the roll direction of the current time's gait (around the X axis) As shown in FIG. 28, the angular trajectory is preferably set so as to smoothly change from “0” to the one-leg hopping roll direction body posture angle.

  In the embodiment described above, after determining the normal gait as a virtual periodic gait following the current time's gait, the terminal divergence component of the current time's gait is substantially matched with the terminal divergence component of the normal gait. On the other hand (so that the current time's gait can converge to a normal gait in the future), the current time's gait is generated. However, the present invention does not necessarily determine the normal gait. That is, as long as it is a technique that can generate a gait that can maintain the continuous stability of the posture of the robot 1, a gait such as a one-leg hopping gait is generated by other techniques. Also good.

  Further, the dynamic model for generating gaits is not limited to the dynamic model shown in FIG. For example, a multi-mass model having a mass on each link of the robot 1 may be used as the dynamic model. As the gait generation dynamic model, any other dynamic model may be used as long as it can appropriately express the dynamics of the robot 1 in a one-leg hopping motion or the like.

  Alternatively, the target gait is generated after the target gait is generated using a relatively simple dynamic model, such as the gait generation technique disclosed by the present applicant in, for example, Japanese Patent No. 4246538. May be generated using a stricter dynamic model to generate various desired gaits including a one-leg hopping gait.

  In the above-described embodiment, the case where the target gait is generated in real time during the operation of the robot 1 has been described as an example. The robot may generate a target gait of the robot at any time using a computer.

  DESCRIPTION OF SYMBOLS 1 ... Biped mobile robot, 2R, 2L ... Leg, 24 ... Upper body, 26 ... Control unit (target gait generating means, target motion generating means, joint control means), 100 ... Gait generating device (target gait) Generating means, target motion generating means).

Claims (6)

A bipedal mobile robot having two legs whose base ends are connected to the upper body via joints. As an operation that can be realized by the robot, one of the two legs is removed from the floor. A control device for a biped mobile robot including a one-leg hopping operation in which a floor leaving operation and a subsequent landing operation are performed only by the other leg in the state where
When the robot performs the one-leg hopping operation, the base end of the one leg is more than the base end of the other leg in the landing state after leaving the other leg. Target gait generating means for generating a target gait including at least a target motion of the robot so as to exist at a relatively high position;
A biped mobile robot control device comprising: joint control means for controlling the operation of each joint of the robot according to the generated target gait. A bipedal mobile robot having two legs whose base ends are connected to the upper body via joints. As an operation that can be realized by the robot, one of the two legs is removed from the floor. A control device for a biped mobile robot including a one-leg hopping operation in which a floor leaving operation and a subsequent landing operation are performed only by the other leg in the state where
When causing the robot to perform the one-leg hopping operation, the position of the center of gravity of the entire robot is set to the position of the center of gravity of the robot and the other leg in the landing state after leaving the other leg. A target that generates a target gait that includes at least the target motion of the robot so that a horizontal distance to the base end of the body is shorter than the horizontal distance in a state where the robot stands in an upright posture Gait generating means;
A biped mobile robot control device comprising: joint control means for controlling the operation of each joint of the robot according to the generated target gait. The control apparatus for a biped mobile robot according to claim 1 or 2,
The target motion generated by the target gait generating means in the landing state after leaving the other leg is that the center of gravity of the entire robot and the base end of the other leg are A control device for a biped mobile robot, characterized in that the target motion is arranged so as to be arranged vertically above the ground contact surface of the leg. A bipedal mobile robot having two legs whose base ends are connected to the upper body via joints. As an operation that can be realized by the robot, one of the two legs is removed from the floor. A gait generator for a biped mobile robot including a one-leg hopping operation in which a floor leaving operation and a subsequent landing operation are performed only by the other leg in the state where
When generating a target gait that includes at least the target motion of the robot for causing the robot to perform the one-leg hopping operation, the base of the other leg in the landing state after leaving the other leg A bipedal mobile robot comprising target motion generation means for generating a target motion of the robot so that a base end portion of the one leg is located at a relatively higher position than an end portion. Gait generator. A bipedal mobile robot having two legs whose base ends are connected to the upper body via joints. As an operation that can be realized by the robot, one of the two legs is removed from the floor. A gait generator for a biped mobile robot including a one-leg hopping operation in which a floor leaving operation and a subsequent landing operation are performed only by the other leg in the state where
When generating a target gait that includes at least the target motion of the robot for causing the robot to perform the one-leg hopping operation, the center of gravity of the entire robot in the landing state after leaving the other leg A target motion that generates a target motion of the robot such that a horizontal distance between the base and the base end of the other leg is shorter than the horizontal distance when the robot stands upright A gait generating device for a biped mobile robot, characterized by comprising generating means. The gait generator for a biped mobile robot according to claim 4 or 5,
The target motion generated by the target gait generating means in the landing state after leaving the other leg is that the center of gravity of the entire robot and the base end of the other leg are 2. A gait generator for a biped mobile robot, wherein the gait generator is a target motion arranged so as to be lined up and down immediately above the ground contact surface of the leg.

Images