JP5633166B2 - Robot and control method thereof - Google Patents

Description

  The present invention relates to a robot and a control method thereof, and more particularly to a robot that guides a person to be guided and a control method thereof.

  In a legged mobile robot, a technique for walking while holding hands with a person is disclosed (Patent Document 1). The mobile robot of Patent Document 1 detects an external force acting from a person when contacting the person via a hand, and generates a gait based on the detected external force. In addition, compliance control is performed so as to absorb disturbance of external force. Furthermore, the hand position in the direction of gravity is determined according to the height of the person. Then, the robot walks with the robot holding hands with people. In this way, the robot guides people.

JP 2007-185663 A

  By the way, it is preferable that a humanoid robot can be guided to a destination by pulling the hand of a visually impaired person or an elderly person. In this case, it is necessary to control in accordance with how the person follows following the safety of the person being guided. For example, when guiding a person, it is necessary to pull the person in the guiding direction. Furthermore, the walking speed may vary depending on the person. The above-mentioned Patent Document 1 does not disclose any method for performing control from such a viewpoint, and thus there is a problem that it cannot be guided appropriately.

  The present invention has been made in view of the above-described problems, and an object thereof is to provide a robot that can be guided appropriately and a control method thereof.

  A robot according to a first aspect of the present invention is a robot that has an arm portion having a hand, a leg portion, and a torso portion, and moves by guiding a guided person from a movement start point to a movement end point. A gait data generation unit that generates gait data according to the travel route to the hand, a hand position calculation unit that calculates a target position of the hand at a predetermined robot position based on the gait data, and the predetermined A tracking state estimation unit that estimates the tracking state of the guided person with respect to the robot according to a deviation between the target position of the hand at the robot position and the measurement position of the hand, and an estimation result of the tracking state estimation unit An adjustment unit that adjusts the movement speed of the robot according to the control unit, and a drive control unit that drives the leg unit to move the robot at the movement speed adjusted by the adjustment unit. That.

  A robot according to a second aspect of the present invention is the robot described above, wherein the robot moves by changing the stride length of the leg according to a deviation between the target position of the hand and the measurement position. It changes the speed.

  A robot according to a third aspect of the present invention is the robot described above, wherein a position with a foot after changing the stride is calculated along the movement path.

  A robot according to a fourth aspect of the present invention is the robot as described above, wherein the stepped foot after changing the stride is on a straight line passing through a virtual stepped position when the free leg lands at the reference step length. The position is to be determined.

  A robot according to a fifth aspect of the present invention is the robot described above, wherein the stride changes linearly according to the deviation.

  A robot according to a sixth aspect of the present invention is the above robot, wherein when the hand position deviation is smaller than a threshold value, the position with the foot after changing the stride is the position with the foot of the support leg. The stride is adjusted so that it matches.

  A robot according to a seventh aspect of the present invention is the robot described above, wherein the hand position calculation unit is configured such that a lateral variation amount of the target position of the hand with respect to the movement path is a position of the body part with respect to the movement path. The target position of the hand is calculated so as to be smaller than the lateral variation amount.

  A robot according to an eighth aspect of the present invention is the robot described above, and impedance-controls the position of the hand according to a sensor signal from a sensor unit provided in the hand and a target position of the hand. And an impedance control unit.

  A robot according to a ninth aspect of the present invention is the above robot, wherein the impedance control pulls the guided person in a traveling direction along the moving path, and the traveling direction is the traveling direction. The impedance control is performed so as to be softer than the orthogonal direction.

  A robot according to a tenth aspect of the present invention is the robot described above, and detects that the guider and the robot are in a non-contact state based on a sensor signal from a sensor unit provided in the hand. In this case, the movement and stop by the legs are stopped.

  A robot according to an eleventh aspect of the present invention is the robot described above, wherein the movement locus is determined in consideration of a sum of widths of the robot and the guided person. It is.

  A robot control method according to a twelfth aspect of the present invention is a robot control method that includes an arm portion having a hand, a leg portion, and a torso portion, and guides and moves the guided person. Generating gait data according to a movement route from a movement start point to a movement end point; calculating a target position of the hand at a predetermined robot position based on the gait data; and Estimating a follow-up state of the guided person with respect to the robot according to a deviation between the target position of the hand at the robot position and the measurement position of the hand; Adjusting the moving speed, and driving the leg to move the robot at the adjusted moving speed.

  A robot control method according to a thirteenth aspect of the present invention is the control method described above, wherein the step length of the leg is changed in accordance with a deviation between the target position of the hand and the measurement position. It changes the moving speed of the robot.

  A robot control method according to a fourteenth aspect of the present invention is the control method described above, wherein a position with a foot after changing the stride is calculated along the movement route.

  A robot control method according to a fifteenth aspect of the present invention is the control method described above, wherein the stride is changed on a straight line passing through a virtual legged position when the free leg lands at a reference stride. The position with the subsequent foot is determined.

  A robot control method according to a sixteenth aspect of the present invention is the control method described above, wherein the stride changes linearly according to the deviation.

  The robot control method according to a seventeenth aspect of the present invention is the control method described above, wherein when the hand position deviation is smaller than a threshold value, the position with the foot after changing the stride is a support leg. The stride is adjusted so that it is aligned with the position of the foot.

  A robot control method according to an eighteenth aspect of the present invention is the control method described above, wherein a lateral variation amount of the target position of the hand with respect to the movement path is a lateral direction of the body part position with respect to the movement path. The target position of the hand is calculated so as to be smaller than the fluctuation amount.

  A robot control method according to a nineteenth aspect of the present invention is the control method described above, wherein the position of the hand is determined according to a sensor signal from a sensor unit provided in the hand and a target position of the hand. Is for impedance control.

  A robot control method according to a twentieth aspect of the present invention is the control method described above, wherein the impedance control pulls the guided person in a traveling direction along the movement path, and the traveling direction is The impedance is controlled so as to be softer than the direction orthogonal to the traveling direction.

  A robot control method according to a twenty-first aspect of the present invention is the control method described above, wherein the guider and the robot are brought into a non-contact state by a sensor signal from a sensor unit provided in the hand. When it is detected that there is a movement, the movement by the leg portion is stopped.

  A robot control method according to a twenty-second aspect of the present invention is the control method described above, wherein the movement trajectory is determined in consideration of a sum of widths of the robot and the guided person. It is a feature.

  ADVANTAGE OF THE INVENTION According to this invention, the robot which can be guide | induced appropriately, and its control method can be provided.

It is a figure which shows typically the whole structure of the robot concerning this invention. It is a block diagram which shows the structure of the control part of the robot concerning this invention. 4 is a flowchart showing a robot control method according to the present invention. It is a figure which shows the movement path | route of the robot concerning this invention. It is a figure which shows the movement path | route of the robot concerning this invention, and the movement path | route of a person. It is a figure which shows the movement path | route and hand position track | orbit of a robot concerning this invention. It is a top view which shows the coordinate system centering on the hand position of the robot concerning this invention. It is a perspective view which shows the coordinate system centering on the hand position of the robot concerning this invention. It is a figure which shows a mode that abnormality occurred in the following state in the robot concerning this invention. It is a figure which shows a mode that abnormality occurred in the following state in the robot concerning this invention. It is a figure for demonstrating the stride adjustment process of the robot concerning Embodiment 1 of this invention. 6 is a flowchart showing processing for adjusting a stride in the robot according to the embodiment of the present invention. It is a flowchart which shows the induction | guidance | derivation operation | movement which considered the control period. It is a figure which shows a mode that a stride is adjusted linearly. FIG. 10 is a diagram for explaining a stride adjustment process of the robot according to the second exemplary embodiment; FIG. 10 is a top view showing the stride adjusted by the robot control method according to the second embodiment;

  Hereinafter, embodiments of a moving body according to the present invention will be described in detail with reference to the drawings. However, the present invention is not limited to the following embodiments. In addition, for clarity of explanation, the following description and drawings are simplified as appropriate.

Embodiment 1 FIG.
Hereinafter, a legged mobile robot (hereinafter simply referred to as a robot) according to a first embodiment of the present invention will be described with reference to FIG. FIG. 1 is a front view schematically showing the overall configuration of the robot. The robot according to the present embodiment is a humanoid robot that pulls and guides a human hand. That is, when the robot 1 pulls a person and walks, the person follows the robot.

  FIG. 1 is a schematic view schematically showing a state in which the robot 1 is viewed from the front, and shows a state in which the robot 1 walks on the floor F. In FIG. The x direction is the direction in which the robot travels (front-rear direction), the y-axis is the direction orthogonal to the horizontal direction (left-right direction) in the direction in which the robot 1 travels, and the direction (vertical direction) extends in the vertical direction from the plane on which the moving body moves. A description will be given using the coordinate system consisting of these three axes as the z-axis. That is, in FIG. 1, the x-axis indicates the depth direction of the paper surface, the y-axis indicates the left-right direction toward the paper surface, and the z-axis indicates the vertical direction in the paper surface. ,

  As shown in FIG. 1, the robot 1 has a head 2, a trunk 3, a waist 4 coupled to the trunk 3, a right arm 5, a left arm 6, and a waist 4 connected to the trunk 3. And a leg unit 10 that is rotatably fixed. Details will be described below.

  The head 2 includes a pair of left and right imaging units (not shown) for visually imaging the environment around the robot 1, and the head 2 is parallel to the trunk 3 in the vertical direction. By rotating around the axis, the surrounding environment is imaged widely. Image data indicating the captured surrounding environment is transmitted to the control unit 130 described later, and is used as information for determining the operation of the robot 1.

  The trunk 3 houses therein a control unit 130 that controls the operation of the robot 1 and a battery (not shown) for supplying power to actuators (for example, motors) of legs and arms. It is. As will be described later, the control unit 130 drives the leg unit 10 and stores a gait data for moving the robot 1, a calculation processing unit for reading out the gait data stored in the storage region, A drive control unit that drives a motor included in the leg unit 10. Each of these components operates by being supplied with electric power from a battery (not shown) provided inside the trunk 3.

  The arithmetic processing unit reads out the gait data stored in the storage area and calculates the joint angle of the leg 10 necessary to realize the posture of the robot 1 specified by the read out gait data. Then, a signal based on the joint angle calculated in this way is transmitted to the drive control unit. Known techniques can be used for the generation of gait data and the basic control of walking based on gait data.

  The drive control unit specifies the drive amount of each motor for driving the leg portion based on the signal transmitted from the arithmetic processing unit, and outputs the motor drive signal for driving the motor with these drive amount to each motor. Send to. As a result, the driving amount at each joint of the leg 10 is changed, and the movement of the robot 1 is controlled.

  In addition to instructing the motor to be driven based on the read gait data, the arithmetic processing unit receives a signal from a sensor (not shown) such as a gyroscope or an accelerometer built in the robot 1, Adjust the drive amount of the motor. Further, a laser sensor or the like may be provided to detect the distance to the floor surface F or the distance to the obstacle. Various sensors such as a gyro sensor, an accelerometer, and a laser sensor are provided on the trunk 3 and the waist 4, for example. Thus, the robot 1 can be maintained in a stable state by adjusting the joint angle of the leg 10 according to the external force acting on the robot 1 detected by the sensor, the posture of the robot 1, and the like.

  The right arm 5 and the left arm 6 are connected to the trunk 3 so as to be rotatable, and by driving joint portions provided on the shoulder portion, the elbow portion, and the wrist portion, the same movement as that of a human arm portion is performed. It can be performed. Further, the hand part connected to the tip of the wrist part has a hand structure for gripping an object (not shown), and by driving a plurality of finger joints incorporated in the hand structure, various kinds of parts are provided. It becomes possible to grip an object having a shape. For example, by driving a finger joint, a guiding person's hand can be grasped. When guiding a person, the robot 1 may hold the person's hand, or the person may hold the hand of the robot 1. Furthermore, the robot 1 and the person may hold each other's hands and hands.

  The lumbar part 4 is connected to the trunk 3 so as to rotate, and the driving energy necessary for driving the leg part 10 can be obtained by combining the rotational action of the lumbar part 4 when performing a walking action. Can be reduced. The trunk 3 and the waist 4 constitute the torso part.

  A leg 10 (right leg 20 and left leg 30) for bipedal walking is composed of a right leg 20 and a left leg 30. Specifically, as shown in FIG. 2, the right leg 20 includes a right hip joint 21, an upper right thigh 22, a right knee joint 23, a right lower leg 24, a right ankle joint 25, and a right foot 26. A left hip joint 31, a left upper thigh 32, a left knee joint 33, a left lower thigh 34, a left ankle joint 35, and a left foot 36 are provided.

  The right leg 20 and the left leg 30 are each driven to a desired angle by transmitting a driving force from a motor (not shown) via a pulley and a belt (not shown). A desired movement can be made to a leg part.

  In this embodiment, when the leg 10 (the right leg 20 and the left leg 30) lifts the lower leg to the front side around the knee joint, the upper leg and the lower leg are placed rearward like a human leg. It is in an open state (a state where the lower leg rotates around the knee joint on the rear side of the extension line of the upper leg).

  The gait data stored in the storage area is associated with the amount of movement of the leg 10, the position of the tip (foot tip) of the foot (the right foot 26, the left foot 36) of the leg 10, the mobile body This position is indicated over time in a coordinate system (for example, an xyz coordinate system) that defines a space in which the robot 1 moves.

  Next, the arm part provided in order to pull a person's hand is demonstrated concretely. The right arm 5 and the left arm 6 are arm mechanisms having a plurality of degrees of freedom, for example, and can drive the hand (hand) to a desired position. Specifically, an arm mechanism having 7 or 8 degrees of freedom can be used. Of course, an arm mechanism having other degrees of freedom may be used. A driving force from a motor (not shown) is transmitted via a pulley and a belt (not shown), so that each joint portion is driven at a desired angle, and as a result, the leg portion can be moved in a desired manner.

  The right arm 5 includes a right hand 51, a right wrist joint 52, a right forearm 53, a right elbow joint 54, an upper right arm 55, and a right shoulder joint 56. The right wrist joint 52, the right elbow joint, and the right shoulder joint 56 have a rotary joint mechanism and the like. By controlling these joints by the control unit 130, the right arm 5 can set the hand position of the right hand 51 to a desired position. Further, the right hand 51 is provided with a right hand sensor unit 57.

  Similarly, the left arm 6 has a left hand 61, a left wrist joint 62, a left forearm 63, a left elbow joint 64, a left upper arm 65, and a left shoulder joint 66. The left wrist joint 62, the left elbow joint 64, and the left shoulder joint 66 have a rotary joint mechanism and the like. By controlling these joints by the control unit, the left arm 6 can set the hand position of the left hand 61 to a desired position. Further, the left hand 61 is provided with a left hand sensor 67. In addition, when always guiding with the same arm, the sensor part should just be provided only in one arm. Hereinafter, one of the right hand sensor unit 57 and the left hand sensor unit 67 is referred to as a hand sensor unit, and unless otherwise noted, the hand sensor unit provided on the hand that is pulling the person's hand. Point to.

  The control unit 130 is an arithmetic processing unit having a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), a communication interface, and the like. The control unit 130 includes a removable HDD, optical disk, magneto-optical disk, and the like, stores various programs, control parameters, and the like, and supplies the programs and data to a memory (not shown) or the like as necessary. . Of course, the control unit 130 is not limited to one physical configuration physically. The control unit 130 performs arithmetic processing for controlling the leg 10, the right arm 5, and the left arm 6. That is, a command value is output to an actuator (for example, a motor) of each joint.

  The right hand sensor unit 57 and the left hand sensor unit 67 have one or more sensors for estimating the user's follow-up state. For example, the right hand sensor unit 57 and the left hand sensor unit 67 include a touch sensor, a force sensor, a contact sensor, and the like. For example, it is possible to determine whether or not the user's hand is being pulled by providing a touch sensor or a contact sensor on the palm portion of the robot hand. Furthermore, the force sensor detects the force applied by the person. The force sensor may differentiate the influence of its own weight.

  Control using the sensor signal output from the hand sensor unit will be described with reference to FIGS. FIG. 2 is a block diagram illustrating a configuration of the control unit 130. FIG. 3 is a flowchart illustrating a control method by the control unit 130. The robot 1 is guiding a person (guided person) with a hand. Therefore, the control unit 130 performs control according to the sensor signal output from the hand sensor unit provided in one hand. In the following description, it is assumed that the left hand 61 is guiding while holding a human hand.

  The control unit 130 includes a route calculation unit 131, a gait data calculation unit 132, a hand position calculation unit 133, an impedance control unit 134, a contact state determination unit 135, a tracking state estimation unit 136, a walking adjustment unit 137, and a drive control unit 138. I have. Further, a sensor signal from the left hand sensor unit 67 is input to the control unit 130. In FIG. 2, the output value of the encoder of the joint motor is omitted.

  First, a robot route for guiding a person is planned (step S101). First, the route calculation unit 131 calculates a movement route from the movement start point to the movement end point. The route calculation unit 131 detects from the sensor signal from the left hand sensor unit 67 that the left hand 61 has grasped the person's hand to be guided. Specifically, contact is detected by a sensor signal from a contact sensor, a touch sensor, or the like. As a result, it is recognized that the guidance is possible. Then, a movement route from the current position (movement start point) to the input target position (movement end point) is calculated. Further, the position of the obstacle is measured by a laser sensor, a camera, or the like, and the movement path is calculated so as to avoid the sparse obstacle. Note that the movement path may be changed as appropriate according to the measurement results of the laser sensor, the camera, or the like. That is, when an obstacle moves, a new movement route may be calculated so as to avoid the obstacle.

  Further, the movement route may be updated according to the change in the position of the obstacle 200. For example, an obstacle is detected by a camera or a laser sensor provided in the robot 1. Moreover, you may detect an obstruction with the camera installed in the moving environment. And when the position of an obstacle changes, you may change a movement route. That is, when the obstacle moves to the vicinity of the movement path and there is a possibility that the robot 1 and the person 201 are prevented from moving, the movement path is changed.

  The gait data calculation unit 132 calculates gait data on the movement route. The gait data refers to the position of the tip (foot tip) of the foot of the leg 10 (the right foot 26, the left foot 36) and the position of the robot 1 body in association with the amount of movement of the leg 10. This is instructed over time in a coordinate system (for example, an xyz coordinate system) that defines a space in which the robot 1 moves. In addition, about the production | generation of gait data, since the method widely used conventionally can be used, description is abbreviate | omitted. As the right foot 26 and the left foot 36 land alternately, the robot 1 walks. For this reason, the gait data includes the position where the foot lands (position with a foot). In this way, the route calculation unit 131 and the gait data calculation unit 132 execute the route plan to the movement end point. That is, a foot plan for walking with both legs is planned. Note that the gait data need not calculate gait data for the entire travel route. For example, the gait data may be calculated only for a part of the route close to the movement start point, and the gait data may be calculated at any time during walking.

  Next, restraint control is performed on the hand position (hand position) during walking (step S102). Therefore, the hand position calculation unit 133 calculates a target hand position based on the gait data. Here, the hand position of the left hand 61 holding the user's hand is calculated. When the main body of the robot 1 is at a certain position, the hand position is determined by the joint angle of the arms (the right arm 5 and the left arm 6). For example, the arm joint angle at a certain robot position on the movement path is determined by the swing width of the hand while walking. The hand position is calculated based on the angle of each joint and the length of the link connecting the joints. In this way, the trajectory of the hand position along the movement route is calculated based on the gait data. And the hand position during walking is restrained.

  Then, the impedance of the hand is controlled so that the hand is pulled with a constant force in the traveling direction (step S103). For example, the impedance control unit 134 performs impedance control of the hand position according to a target hand position and a sensor signal from a force sensor or the like. Specifically, the force received by the left hand 61 from a person is detected by a force sensor provided in the left hand sensor 67. Impedance is controlled according to this force. An arm command value for controlling the left arm 6 holding a person is output. By inputting this arm command value to the joint motor of the left arm 6, the position of the hand is controlled. Thereby, the joint motors of the left wrist joint 62, the left elbow joint 64, and the left upper arm 65 are driven at a desired rotational torque and rotational speed. Therefore, a human hand can be pulled with an appropriate force.

  Next, the contact state of the hand and the reaction force from the person are detected (step S104). The force sensor of the left hand sensor unit 67 detects the reaction force. Further, the contact sensor or touch sensor of the left hand sensor unit 67 detects whether or not a person is in contact with the left hand 61 on the robot 1. The contact state determination unit 135 determines the contact state between the person and the left hand 61 of the robot 1 in accordance with these sensor signals. Thereby, it can be discriminate | determined whether it is a non-contact state or a contact state. Then, the tracking state estimation unit 136 estimates the tracking state of the person based on the impedance control result of the hand, the determination result of the contact state determination unit 135, and the like.

  First, it is determined whether or not there is an abnormality in the following state (step S105). The follow-up state estimation unit 136 estimates, for example, a state where a human hand is separated from the hand and is not guided as an abnormal state. Furthermore, the follow-up state estimation unit 136 estimates that an abnormal state occurs when the force received by the robot 1 from a person exceeds a threshold value. In such a case (YES in step S105), the movement is stopped (step S106). For example, the driving of the free leg is stopped when the driving of the supporting leg is stopped and the free leg is landed.

  When not in an abnormal state, the tracking state estimation unit 136 estimates whether the tracking speed is fast or the tracking speed is slow. In other words, the follow-up state estimation unit 136 calculates the difference between the walking speed of the person and the walking speed of the robot according to the target hand position and the impedance control result. When the tracking speed is fast or slow, the next step is calculated by adjusting the stride and the walking cycle according to the deviation from the reference position of the hand (step S106). The walking adjustment unit 137 performs walking adjustment according to the estimation result of the tracking state. For example, when the tracking speed of the person is high, the walking speed of the robot 1 is increased, and when the tracking speed of the person is low, the walking speed of the robot 1 is decreased. Therefore, the walking adjustment unit 137 adjusts the stride and the walking cycle.

  And the drive control part 138 controls the drive of a leg part according to the result adjusted in the walk adjustment part 137. FIG. That is, the walking command value is output to the leg joint motor. Thereby, the joint motors of the right hip joint 21, the left hip joint 31, the right knee joint 23, the left knee joint 33, the right ankle joint 25, and the left ankle joint 35 are driven at a desired torque and rotational speed. Therefore, the robot 1 walks in a desired direction along the movement path at an appropriate speed.

  It is determined whether or not the target position, that is, the movement end point has been reached (step S108). When the target position is reached, walking is stopped (step S109). That is, the drive of the leg motor is stopped. Further, the force for guiding the person with the left hand is set to zero. On the other hand, if the target position has not been reached, the processing from step S102 is repeated. The walking is continued until the target position is reached or until an abnormal state is estimated. In this way, walking control is performed. The above process will be described in detail below.

(Route planning)
The route plan will be described with reference to FIGS. 4 and 5. FIG. 4 is a diagram schematically showing the movement route. FIG. 5 is a diagram illustrating a method for calculating a movement route that avoids an obstacle. Hereinafter, processing by the route calculation unit 131 will be described.

  Here, the travel route can be calculated using various known methods. As shown in FIG. 4, in order to avoid the obstacle 200, a movement route R from the movement start point S to the movement end point G of the movement route is calculated. For example, RRT (Rapidly Exploring Random Trees) that is a probabilistic search method or an A * algorithm is used to plan paths discretely. Then, each discrete point is set as a subgoal SG, and the space between the subgoals is complemented with a spline curve, a Pezier curve, a linear shape, or the like. Thereby, a trajectory of the movement route is generated. Furthermore, map information of the moving environment stored in advance by the robot 1 may be used for generating the moving route. Furthermore, you may utilize the information of the obstruction acquired with the camera or the laser sensor.

  Further, at the time of route planning, as shown in FIG. 5, the moving route is calculated with a margin in consideration of the width of the person. Consider the distance (Lr) from the center point of the robot 1 to the hand position of the robot 1 on the side that pulls the person 201, and the width (Lh) of the person 201. Then, the travel route is calculated with a margin of the sum Lall of the distance Lr and the width Lh. That is, the distance from the moving route to the obstacle 200 is set to be longer than Lall. Thereby, the person 201 can walk so as not to interfere with the obstacle. Of course, it is not necessary to consider the width Lh for the hand opposite to the person 201. Thus, the movement route is calculated with a margin so that the guided person 201 can avoid the obstacle 200. As a result, a person can walk without being aware of the obstacle 200. It is preferable that the person width Lh is set in advance. Of course, the width Lh of the person may be changed according to the person 201 to be guided. The total width of the robot 1 and the person 201 is 2Lr + Lh, and the movement route R through which this width can pass is calculated.

  Then, the gait data calculating unit 132 calculates gait data so as to walk along the moving route R. Here, the footed positions of both feet are calculated. At this time, the ideal trajectory of the hand position of the robot 1 is Rd, and the trajectory of the human hand (the hand on the side opposite to the robot 1) is Rh.

(Restraint control)
The restraint control of the hand position will be described with reference to FIG. It is a top view which shows the robot 1 and the person 201 which move FIG. 6 according to a movement path | route. Further, in FIG. 6, the locus of the footed position Tr of the right foot 26 and the footed position Tl of the left foot 36 is shown. The footed position indicates a position where each foot touches down and can be calculated according to gait data. The footed position Tr of the right foot 26 is on the right side of the moving route R, and the footed position Tl of the left foot 36 is on the left side of the moving route R. First, a problem when the constraint control is not performed will be described.

  In order to walk stably with a biped walking robot, it is necessary to perform an operation so that the support polygon formed by the support legs contains the center of gravity for static walking and ZMP (Zero Moment Point) for dynamic walking. In general, a biped robot walks according to the above restrictions. For this reason, the waist center is a trajectory Rb with a lateral variation with respect to the traveling direction of the movement path. For example, when the right foot 26 is in contact with the ground, the waist center track Rb is on the right side of the movement route R, and when the left foot 36 is installed, the waist center track Rb is on the left side of the movement route R. As described above, the waist center trajectory Rb during walking is composed of the supporting legs, and therefore the waist center trajectory Rb is accompanied by a lateral variation with respect to the movement route R.

  When restraint control is not performed, the hand position also changes as the waist center trajectory Rb changes in the lateral direction. That is, as the robot 1 walks, the waist center of the robot 1 and the hand position change. Then, during the movement, the hand of the person 201 will shake from side to side. That is, the person 201 receives a force from the robot 1 in accordance with the lateral variation of the waist center. It is not preferable for a person to receive such power. For example, when a person receives a lateral force, the movement direction is erroneously recognized.

  Therefore, in this embodiment, constraint control is performed so that the lateral variation amount of the track Rd at the hand position during movement is smaller than the lateral variation amount of the waist center track Rb. Here, the lateral direction fluctuation amount indicates a variation in the distance between each track and the moving route R. Therefore, if the hand position trajectory Rd is parallel to the movement path R, the distance between the hand position trajectory Rd and the movement path is constant. When the hand position trajectory Rd is parallel to the movement path, the lateral variation amount is zero. On the other hand, the waist center trajectory Rb swings to the left and right so as to cross the movement route R. For this reason, the distance between the waist center trajectory Rb and the movement path changes according to the position on the movement path, and the variation in distance increases. In this way, the hand position of the robot 1 is restrained and controlled so that the hand position trajectory Rd is substantially parallel to the movement path R.

  Even if the waist center position is accompanied by lateral fluctuations, the hand position to be guided is restrained and controlled so that there is no lateral fluctuation. The control unit 130 performs this constraint control in real time. Actually, the trajectory Rd of the hand position is made parallel to the movement path while satisfying the above-mentioned center of gravity and ZMP constraints. That is, the control unit 130 performs control so that the hand position moves with a certain distance from the movement route R. In order to perform this control, the control unit 130 obtains the joint angle of the whole body, for example, by solving inverse kinematics. Inverse kinematics may be solved by an analytical solution or by numerical processing using a Jacobian. Specifically, the joint angles of the whole body may be obtained in consideration of accuracy and calculation load according to the link configuration of the robot 1. As a result, the hand position moves along the movement route R. The trajectory Rd at the hand position is parallel to the movement path R, and the lateral fluctuation amount can be reduced. Therefore, it is possible to avoid a person from erroneously recognizing the guidance direction.

(Impedance control)
The impedance control of the hand position will be described with reference to FIGS. FIG. 7 is a top view showing a coordinate system for clarifying the explanation of the impedance control of the hand position. FIG. 8 is a perspective view showing impedance control of the hand position. Hereinafter, processing by the impedance control unit 134 will be described.

  First, as shown in FIGS. 7 and 8, a right-handed local coordinate system is defined. That is, the moving direction of the robot 1 is the X direction, and the direction perpendicular to the X direction in the horizontal plane is the Y direction. Further, the direction perpendicular to the X direction and the Y direction, that is, the vertical direction is defined as the Z direction. A point on the hand position trajectory is set as the origin. The X direction is parallel to the movement route R at that point. The impedance is controlled according to the following policy. As described above, the hand position used for impedance control is calculated by the hand position calculation unit 133, and the force applied to the left hand 61 is measured by the left hand sensor unit 67.

(A) X axis (traveling direction)
The impedance is controlled so that a force Fr for pulling the person 201 can be presented in the traveling direction. Specifically, a force that pulls the person 201 with a constant light force (Fr: about 5 to 10 N) is generated. Further, when the person 201 applies a reaction force (about 5 to 20 N) in the reverse direction, the control is performed with a softness that can follow the reverse direction.

(B) Y axis (lateral direction)
Impedance is controlled to a certain degree of hardness against human reaction force. For example, when the person 201 applies a reaction force (a load of about 20 N) to the left hand, the impedance is controlled so as to follow about 1 to 2 cm. That is, when the person 201 pulls the left hand 61 with a force of about 20 N, the left hand 61 is pulled about 1 to 2 cm.

(C) Z axis (vertical direction)
Impedance is controlled to a certain degree of hardness against human reaction force. For example, when the person 201 applies a reaction force (a load of about 10 N) to the left hand, the impedance is controlled so as to follow about 3 to 4 cm. That is, when the person 201 pulls the left hand 61 with a force of about 10 N, the left hand 61 is pulled about 3 to 4 cm.

(D) Rz axis (rotation around the Z axis: rotation in the yaw direction)
Even when the robot 1 turns and the direction of travel changes, impedance control is performed so that the robot 1 can follow the human hand to some extent.

(E) Rx axis (rotation around X axis: rotation in roll direction), Ry axis (rotation around Y axis: rotation in pitch direction)
It is very hard to control and it is not necessary to follow the external force. Therefore, impedance control is not necessary for the Rx axis and the Ry axis.

  Thus, impedance control is not performed on the Rx axis and the Ry axis. Furthermore, the X axis is controlled to be the softest and with a constant force. Further, the Y axis is controlled to be harder than the X axis, the Z axis, and the Rz axis. Therefore, the softness of the impedance control is in the following order. X axis {softest} → Rz axis, Z axis → Y axis → (Rx axis, Ry axis)

  As an application example for realizing the above impedance control, a control law such as the following equation (1) can be used.

  In Equation (1), p is a vector of the current hand position (x, y, z, Rx, Ry, Rz), pd is a hand position command (target hand position), M is a real inertia matrix, and B is a damper. Matrix, K is a virtual spring matrix, F is a force measurement value, Fd is a force command value, S is a switch matrix, Kf is a force gain, and E is a unit vector. The hand position command pd corresponds to the target hand position calculated in the constraint control.

  By manipulating the switch matrix in the above equation (1), not only the hand can be controlled compliantly but also a force corresponding to an arbitrary force command value can be generated. Therefore, the switch matrix is set so that the force command value is valid only in the X direction, which is the moving direction, and compliant control is performed in the other directions. Regarding the control hardness in each direction, parameters of the inertia matrix, the damper matrix, and the spring matrix are set so as to realize the desired hardness shown in (a) to (e) above.

  The target hand position is calculated by constraint control. Then, by calculating the force command value by combining the mass element, the damper element, and the spring element from the target hand position, guidance with an appropriate force becomes possible. Then, according to the impedance control result, the impedance control unit 134 outputs an arm command value. Thereby, the hand position can be driven according to the impedance control result. In consideration of the movable range of the arm and the self-interference of the robot, control is performed so as to keep the previous arm position when the robot cannot move any more. In this manner, the impedance control unit 134 pulls a person in the traveling direction along the movement path, and performs impedance control so that the X direction is softer than the Y direction and the Z direction orthogonal to the X direction.

(Abnormal state control)
The control when it is determined that the following state is an abnormal state will be described. 9 and 10 are diagrams for explaining an example in which a tracking abnormality has occurred. As shown in FIG. 9, when the hand of the person 201 moves away from the left hand 61, there is no reaction at the touch sensor, contact sensor, or force sensor of the left hand sensor unit 67. Accordingly, it is possible to determine whether or not the hand of the person 201 has been released based on the sensor signal from the left hand sensor unit 67. Hereinafter, processing by the contact state determination unit 135 and the tracking state estimation unit 136 will be described.

  In addition, when using a force sensor, it is preferable to consider the output value due to the influence of its own weight as a difference. Further, when a sensor that outputs an analog value instead of ON / OFF is used, it may be determined whether or not a contact is made by comparing the value of the sensor signal with a certain threshold value. In order to prevent sensor chattering, it is preferable to determine that a person has left when there is no response for several milliseconds or longer. Furthermore, in the touch sensor or the contact sensor, in order to determine whether or not the person 201 is holding the left hand 61 of the robot 1, it is preferable to install sensors on both the back side and the palm side of the left hand 61. . In this case, the determination accuracy can be improved by taking both reactions. For example, when any one of the sensors no longer responds, it is determined that the human hand has left the robot hand. Of course, when both responses are lost, it may be estimated that the human hand has left the robot hand. Note that the various sensors are provided at positions where they are normally contacted when a human hand holds the hand of the robot 1.

  Also, as shown in FIG. 10, when a person is about to fall, an excessive reaction force is applied to the robot hand. In this case, it can be estimated whether it is an abnormal state by using the method shown below. The value of the sensor signal from the force sensor of the left hand sensor unit 67 is compared with a threshold value. For example, when the value of the sensor signal exceeds a threshold value, an abnormal state is estimated. Further, when the frequency response of the force sensor is high and exceeds a threshold value, it is estimated as an abnormal state. In this case, an impulse reaction force such as a collision can be detected. If the hand position after application of impedance control is accompanied by a deviation greater than or equal to the threshold with respect to the reference hand position, an abnormal state may be estimated. In this case, different threshold values are set for the X-axis, Y-axis, Z-axis, Rx-axis, Ry-axis, and Rz-axis, respectively. Thus, by comparing the value of the sensor signal or the deviation of the hand position with the threshold value, it is possible to detect an abnormal state in which an excessive force is applied.

  If it is estimated to be abnormal, stop moving in the following procedure. If the current time is the period when both legs are supported, the robot will stop without performing the next walk. That is, before the legs are released, the driving of the leg motor is stopped. Also, if the current time is the one-leg support period, it will stop when the free leg lands next. That is, the leg motor is driven until the free leg lands. Then, after the free leg has landed, the leg motor is stopped. Further, in order to further improve safety, the left arm impedance control may be compliant in all directions. Thereby, it is possible to prevent the balance of the robot 1 from being lost due to the reaction force. For example, each parameter is changed so that the compliance is improved by using the switch matrix in Equation (1) as a unit matrix. Then, control is continued with the changed parameter. As a result, the drive control unit 138 outputs a command value for appropriately stopping the leg motor. Therefore, the robot 1 loses balance and can be prevented from falling.

(Walking adjustment)
Next, a process for adjusting the walking speed according to the estimated tracking state will be described. The walking adjustment unit 137 adjusts the walking speed according to the tracking state estimated by the tracking state estimation unit 136. That is, in step 107, the next step is calculated by adjusting the stride and the walking cycle. In the following processing, processing for adjusting the stride in the walking adjustment unit will be described with reference to FIG.

  FIG. 11 is a top view showing the adjusted stride. Here, Tr indicates the position with the foot of the right foot 26, and Tl indicates the position with the foot of the left foot 36. Further, in this description, an example in which the next step of the right leg 20 is calculated in a state where the left leg 30 is a support leg and the right leg 20 is a free leg will be described. That is, the next footed position of 26 on the right foot is determined. Here, assuming that the reference stride of the robot 1 is Sr, the footed position of the right foot 26 when walking at that stride is Trn. The reference stride Sr can be determined according to the configuration of the robot, the link length, and the reference walking speed.

As shown in FIG. 11, the virtual position of the right foot 26 when both feet are virtually aligned with reference to the left leg 30 as the support leg is defined as Tr ′. A straight line connecting the center of the virtual position Tr ′ of the right foot 26 and the center of the left foot 36 is perpendicular to the movement route R. Here, the next footed position Trn viewed from the virtual position Tr ′ can be expressed as (Srx, Sry, θr) when expressed in (x-direction stride, y-direction stride, turning angle). Here, the next footed position Trn is a position when walking from the virtual position Tr ′ with the reference stride Sr. If the standard stride Sr is expressed using Srx and Sry, Sr = (Srx ² + Sry ² ) ^1/2 .

  As described above, the hand position during walking is impedance controlled. The stride is adjusted in accordance with the deviation (hereinafter referred to as hand position deviation) between the target hand position based on a certain position on the movement path (reference robot position) and the hand position after impedance control. That is, the hand position deviation between the target hand position at the reference robot position and the actually measured hand position is calculated in the previous step. Note that the hand position calculation unit 133 calculates the target hand position at the reference robot position as described above. The actual hand position at the reference robot position can be calculated from the joint angle of the arm at a certain predetermined robot position on the movement path. The next stride is determined using the difference between the target hand position of the reference robot position and the actual hand position as a hand position deviation. As described above, the hand position deviation is a deviation between the target hand position and the actually measured hand position when the body of the robot 1 is at a predetermined position.

  If the hand position deviation after impedance control is on the traveling direction side, it is estimated that the person 201 is sufficiently following the robot 1. That is, it is estimated that the person 201 is walking faster than the robot 1 because the hand is ahead of the target hand position. In order to make the robot 1 walk faster than planned, the stride is made larger than the reference stride. Thereby, the robot 1 is accelerated and the walking speed is increased.

  If the hand position deviation after the impedance control is opposite to the traveling direction, it is estimated that the person 201 does not sufficiently follow the robot 1. That is, it is estimated that the person 201 is behind the robot 1 because the hand is behind the target hand position. In this case, in order to make the robot 1 walk slower than planned, the stride is made smaller than the reference stride. Thereby, the robot 1 decelerates and the walking speed becomes slow.

  An example of the above stride adjustment will be described. Consider the hand position deviation (p-pd) after applying impedance control. Here, only the x component in the formula (1) is extracted. That is, when the vector p is considered only for the x component, the following equation (2) is obtained.

  Thus, the hand position deviation in the traveling direction is represented by (x−xd). Note that x is a measured hand position, and xd is a target hand position. Depending on the value of the deviation (x−xd), the stride is changed according to the following policy.

  For example, if the hand position deviation (x−xd) in the traveling direction is positive, the stride is made larger than the reference stride. That is, when the person is walking faster than the walking speed of the robot 1, the left hand 61 receives a force from the person in the traveling direction. For this reason, the left hand 61 held by a person advances ahead of the target hand position at the reference robot position. Therefore, in order to increase the walking speed of the robot 1, the stride is increased. If the time of one step of the robot 1 is substantially the same, the robot 1 walks faster than the reference speed. On the other hand, if the hand position deviation (x−xd) in the traveling direction is negative, the stride is reduced. That is, when the person is walking slower than the walking speed of the robot 1, the left hand 61 receives a force in the direction opposite to the traveling direction from the person. For this reason, the left hand 61 held by a person is delayed and behind the target hand position at the reference robot position. Therefore, in order to slow down the walking speed of the robot 1, the stride is reduced. If the time of one step of the robot 1 is substantially the same, the robot 1 walks slower than the reference speed.

  If the deviation (x−xd) is smaller than the threshold value Xmin, the stride is set to zero. The threshold value is preferably a negative value. For example, the threshold value Xmin can be set to −0.3 m. Therefore, when the adjusted stride is (Sx, Sy, θ), the stride adjustment can be expressed by the following equations (3) to (6).

When (x−xd) is greater than or equal to Xmin, Sx = (1− (x−xd) / Xmin) × Srx (3)
Sy = (1− (x−xd) / Xmin) × Srx (4)
θ = (1− (x−xd) / Xmin) × θr (5)

When (x−xd) is smaller than Xmin (Sx, Sy, θ) = (0, 0, 0) (6)

  In the above calculation method, when (x−xd) is positive, it can be said that the hand of the person being guided is moving earlier than planned. Therefore, the stride of the robot 1 is increased to accelerate walking. This means that the person can not only guide but also push the robot 1 from the person side to accelerate the movement if it is in the correct traveling direction. Thus, an intuitive guidance operation is realized. In this case, the robot 1 moves according to the walking speed of the person. Therefore, appropriate guidance can be performed. Therefore, it is possible to realize an appropriate guiding operation for various people. Further, the walking cycle may be changed according to the deviation (x−xd) in the traveling direction. In other words, the walking speed may be adjusted by changing the time of one step, that is, the time from taking off to landing after the step length is constant. Of course, the walking speed may be adjusted by changing both the stride and the walking cycle. In general, in biped walking, if the walking cycle becomes too long, stability may be lost. Therefore, it is preferable to perform walking adjustment by changing the stride.

  In the present embodiment, the stride is linearly adjusted according to the hand position deviation (x−xd) in the traveling direction. Therefore, the next footed position changes linearly from the virtual footed position when both feet are aligned with the support leg as a reference. That is, regardless of the hand position deviation, the next footed position is on a certain straight line. This straight line is in the direction connecting the virtual landing position when landing at the standard stride from the virtual leg position of the free leg when both feet are aligned with the support leg as a reference. That is, the next foot position changes on a straight line connecting the virtual position Tr ′ and the foot position Trn at the reference stride. In other words, the footed position footed position after changing the stride exists on one straight line passing through the virtual footed position when the free leg lands at the reference stride.

  When the hand position deviation is greater than or equal to the threshold value, the stride has a linear relationship with the hand position deviation. That is, the distance to the next foot position can be determined according to a value obtained by multiplying the hand position deviation by the reference stride. Thereby, the stride increases as the hand position deviation (x-xd) in the traveling direction increases. Adjustment can be easily performed by linearly adjusting the stride. Further, when the value falls below the negative threshold, control is performed so that both feet are aligned. That is, when a person's walking speed is very slow, both feet are aligned to the left and right. Thereby, it can guide appropriately according to a person's walking speed.

  In the above description, the stride is adjusted focusing on only the X axis, but the stride may be adjusted based on the Y axis and the Z axis. For example, both feet are aligned based on the comparison result between the hand position deviation in the Y-axis direction and the threshold value, or the comparison result between the hand position deviation in the Z-axis direction and the threshold value. Thereby, when an excessive force is applied to the hand on the Y axis and the Z axis, the walking speed can be delayed.

  As described above, when the stride is adjusted, the footed position in the gait data changes. Therefore, it is necessary to update the gait data. Hereinafter, processing for updating gait data will be described with reference to FIG.

  First, route planning is performed first (step S201). Here, the movement route is calculated in the same manner as the processing shown in step S101. That is, the route from the movement start point S to the movement route G is calculated so as not to collide with the obstacle 200. Furthermore, gait data according to the movement route is calculated. Thereby, the coordinates of the footed position along the movement route are calculated.

  Then, position restraint control of the hand while walking is performed (step S202). Here, the target hand position is calculated as in step S102. Then, based on the sensor signal, it is determined whether or not an abnormality has occurred in the following state (step S203). Here, as described above, it is determined whether an abnormality has occurred in the contact state or whether an abnormality has occurred in the reaction force. If the hand of the person 201 is in contact with the left hand 61 and the reaction force received from the person is within a predetermined range, a normal state in which no abnormality has occurred is obtained.

  If no abnormality has occurred (NO in step S203), the hand impedance is controlled (step S204). As a result, the left hand 61 gives an appropriate force to the hand of the person 201. Thereby, the person 201 is pulled in the X direction with a predetermined force. Then, the stride according to the hand position deviation is determined (step S205). That is, as described above, the stride is increased when the hand position is ahead of the target hand position, and the stride is decreased when the hand position is behind the target hand position. Then, walking is continued (step S206). To repeat this process, the process returns to step S201. Here, when the stride is changed from the reference stride Sr, the route is re-planned, and the gait data is adjusted according to the adjusted stride. That is, the footed position changes from the initial position at the adjusted stride. Therefore, gait data corresponding to the new footed position is calculated. In this way, the gait data is updated. Then, the process is repeated in this way to move to the movement start point. Note that the gait data need not be calculated for the entire travel route. That is, when the stride is adjusted, the position with the foot will change, and therefore gait data need only be calculated in the vicinity of the current position of the robot.

  In the above-described process, the hand position constraint control, the impedance control, the walking adjustment, and the update of the gait data may be repeatedly executed at each desired control cycle. Generally, the response is better when the impedance control is performed with a fast control cycle of about several milliseconds. On the other hand, the stride adjustment may be performed when the next step is required in accordance with the walking cycle (for example, 0.5 sec to 2 sec). For example, it is only necessary to determine the next step length before landing. The movement path may be updated by a control period that takes into account the measurement period and calculation speed of the sensor that recognizes the environment. For example, the update cycle of the movement route may be the slowest cycle such as 1 sec to several sec. Considering these control cycles, the flow is as shown in FIG.

  First, information on a sensor that detects an obstacle in the moving environment is acquired (step S301). For example, the position of an obstacle is detected by a camera or a laser sensor. Then, it is determined whether or not the route plan needs to be updated (step S302). In addition, what is necessary is just to perform determination of step S302 with the period of about 1 to several seconds. For example, when sensor information from a camera, a laser sensor, or the like is not sufficiently acquired, the route plan may not be updated. If the position of the obstacle changes and hinders movement, the route plan needs to be updated (YES in step S302).

  When the route plan needs to be updated (YES in step S302), the route plan and the reference walking plan are updated (step S303). As a result, the movement path and the footed position along the movement path are calculated. Thereafter, the hand impedance control is applied (step S304). If it is not necessary to update the route plan (NO in step S302), hand impedance control is applied (step S304). Here, the control period of the impedance control is about several milliseconds. Thereby, a desired force can be generated by the hand.

  It is determined whether the next step is necessary based on the impedance control result (step S305). Here, it is determined whether or not it is time to update the next step. The control period here corresponds to the walking period, and is, for example, 0.5 sec to 2 sec. For example, the next one stride is determined at a certain timing during the one leg support period. Therefore, it is not necessary to determine the next stride until or after that timing. In other words, the next stride may be determined for each step. When the next step is unnecessary (NO in step S305), the joint angle of the whole body is determined (step S307). Thereby, the joint angle of a leg part and an arm part can be calculated | required. If the next step is necessary, the next step is updated according to the latest hand position deviation (step S306). Then, the joint angle of the whole body is determined (step S307). When the joint angle is determined, a command value is output to the actuator corresponding to the joint angle (step S308).

  Thus, the control cycle for updating the next step is made slower than the control cycle for impedance control. For example, the step length of the next step may be determined for each step. Thereby, it can prevent that calculation time becomes long. The hand position deviation may be calculated from the latest robot position. Alternatively, the next stride may be determined based on an average of hand position deviations at a plurality of robot positions.

Embodiment 2. FIG.
The present embodiment is different from the first embodiment in the method of adjusting the stride. Since other processes and the robot configuration are the same as those in the first embodiment, description thereof is omitted. In the first embodiment, the stride is linearly adjusted according to the hand position deviation. However, as shown in FIG. 14, there is a large deviation from the moving route according to the adjustment amount of the stride and the curvature of the moving route. In FIG. 14 , the footed position adjusted linearly with respect to the current footed position Tr and the footed position Trn of the reference stride is indicated by a dotted line. If the position with the foot is greatly deviated from the movement path, for example, the movement path may become longer or the other leg may be interfered. Therefore, when the stride is adjusted by the method according to the present embodiment, the stride can be adjusted more strictly.

As shown in FIG. 15, the current center position of both feet is set as the origin on the movement route. That is, an intermediate position between the center position of the right foot 26 and the center position of the left foot 36 at the adjustment start position is set as the origin (0, 0, 0). Furthermore, the path length is n, and the two-dimensional coordinates and inclinations of points on the path are (fx (n), fy (n), φ (n)). The center position of the right foot corresponding to the point is (xr (n), yr (n), φ (n)), and the center position of the left foot 36 is (xl (n), yl (n), φ (n )). Note that φ (n) is the rotation angle when the counterclockwise direction is positive from the x-axis direction. This φ (n) takes the same angle at the center of both feet, the center of the right foot, and the center of the left foot. If the distance from the center of both feet to the center of the left foot is w, the right foot start position is (0, −w, 0) and the left foot start position is (0, w, 0). When w is used to represent the center position Trn of the right foot 26 and the center position Tln of the left foot 36, the following equations (7) to (9) are obtained.

φ (n) = atan2 (y ′ (n), x ′ (n)) (7)
xr (n) = fx (n) + wsin (φ (n)), yr (n) = fy (n) −wcos (φ (n)) (8)
xl (n) = fx (n) −wsin (φ (n)), yl (n) = fy (n) + wcos (φ (n)) (9)

  And when performing stride adjustment, it adjusts so that the following conditions may be satisfy | filled. The position of the right foot 26 is as shown in the following equations (10) to (13).

When (x−xd) is greater than or equal to Xmin, Sx = xr ((1− (x−xd) / Xmin) × n ′) (10)
Sy = yr ((1− (x−xd) / Xmin) × n ′) + w (11)
θ = φ ((1− (x−xd) / Xmin) × n ′) (12)

When (x−xd) is smaller than Xmin (Sx, Sy, θ) = (0, 0, 0) (13)

  The position of the left foot 36 is as shown in the following formulas (14) to (17).

When (x−xd) is greater than or equal to Xmin, Sx = xl ((1− (x−xd) / Xmin) × n ′) (14)
Sy = yl ((1− (x−xd) / Xmin) × n ′) − w (15)
θ = φ ((1− (x−xd) / Xmin) × n ′) (16)

When (x−xd) is smaller than Xmin (Sx, Sy, θ) = (0, 0, 0) (17)

  In the above formulas (10) to (17), n ′ is the path length at the reference stride. The right foot 26 and the right foot 36 have the same θ. Thus, when adjusting the stride, the position with the foot is changed along the movement route. That is, when adjusting the stride from the hand position deviation of the impedance control result, the next step can be adjusted along the planned movement route. This makes it possible to avoid interference between both feet as shown in FIG. In addition, in FIG. 16, the position with the foot of the right foot 26 when adjusting with various strides is shown. By using the stride adjustment method according to the present embodiment, it is possible to change the position with the foot along the movement route. Furthermore, since the deviation from the movement path is small, the robot 1 can be prevented from making a large turn. Therefore, it is possible to guide at a short distance to the movement end point. Thus, the burden on the person can be reduced.

  By performing impedance control in the robot 1 that pulls a person's hand, the walking direction can be presented intuitively by the force applied to the person. Accordingly, it is possible to appropriately guide to the target movement end point. Furthermore, since the guidance direction is intuitively grasped, it is suitable for appropriately guiding a visually impaired person. Furthermore, the walking speed can be determined in accordance with the person's following state. Thereby, the burden on the to-be-induced person side can be reduced. In addition, when a person's hand leaves or an excessive reaction force occurs, walking is stopped. Thereby, safety can be improved.

  In the above description, the human hand is grasped by the left hand, but the same control can be performed even when the human hand is grasped by the right hand. In the above description, the robot 1 calculates the movement route, but the movement route R may be given in advance. That is, the calculated movement route may be supplied to the robot 1 by another computer or the like. Of course, the stride adjustment may not be performed every step. For example, the stride may be adjusted at a rate of once per a plurality of steps.

  Note that Embodiments 1 and 2 may be combined as appropriate. For example, the stride may be adjusted by the method of the first embodiment in a certain section, and the stride may be adjusted by the method of the second embodiment in another section. The walking control of the robot 1 may be either static walking or dynamic walking. Furthermore, a person may be guided by running with all legs off the floor. Further, the robot 1 according to the present embodiment may be a legged robot other than the bipod type. That is, it may be a legged robot with three or more legs.

DESCRIPTION OF SYMBOLS 1 Robot 2 Head 3 Trunk 4 Lumbar 5 Right arm 6 Left arm 10 Leg part 20 Right leg 21 Right hip joint 22 Right upper leg 23 Right knee joint,
24 right lower leg 25 right ankle joint 26 right foot 30 left leg 31 left hip joint 32 left upper leg 33 left knee joint 34 left lower leg 35 left ankle joint 36 left foot 51 right hand 52 right wrist joint 53 right forearm 54 right elbow joint 55 upper right arm 56 right shoulder joint 57 right hand sensor 61 left hand 62 left wrist joint 63 left forearm 64 left elbow joint 65 left upper arm 66 left shoulder joint 67 left hand sensor 131 path calculation unit 132 gait data calculation unit 133 hand position calculation unit 134 impedance control Unit 135 contact state determination unit 136 tracking state estimation unit 137 walking adjustment unit 138 drive control unit

Claims (20)

An arm having a hand, a leg, and a torso, and a robot that moves by guiding the guided person by pulling the guided person with the hand,
A gait data generation unit that generates gait data according to the movement route from the movement start point to the movement end point;
A hand position calculation unit that calculates a target position of the hand at a predetermined robot position based on the gait data;
A tracking state estimation unit that estimates a tracking state of the guided person with respect to the robot according to a hand position deviation that is a deviation between the target position of the hand and the measurement position of the hand at the predetermined robot position;
An adjustment unit that adjusts the moving speed of the robot according to the estimation result of the tracking state estimation unit;
A drive control unit that drives the leg so as to move the robot at a movement speed adjusted by the adjustment unit ;
The hand position calculation unit calculates the target position of the hand so that a lateral variation amount of the target position of the hand with respect to the movement route is smaller than a lateral variation amount of the body portion position with respect to the movement route. Robot to do. The robot according to claim 1 , further comprising an impedance control unit that performs impedance control on the position of the hand in accordance with a sensor signal from a sensor unit provided on the hand and a target position of the hand. The impedance control unit pulls the guided person in a traveling direction along the moving path, and performs impedance control so that the traveling direction is softer than a direction orthogonal to the traveling direction. Item 3. The robot according to item 2 . The robot according to any one of claims 1 to 3 , wherein a moving speed of the robot is changed by changing a stride of the leg portion in accordance with the hand position deviation. The robot according to claim 4 , wherein the position with the foot after changing the stride is calculated along the movement path. The robot according to claim 4 or 5 , wherein a footed position after changing the stride is determined on a straight line passing through a virtual footed position when the free leg lands at a reference stride. The robot according to claim 6 , wherein the stride changes linearly according to the hand position deviation. The stride length is adjusted according to any one of claims 4 to 7 , wherein when the hand position deviation is smaller than a threshold value, the stride position is adjusted so that a footed position after changing the stride is aligned with a legged position of a support leg. The robot described. By a sensor signal from the sensor unit provided in the hand, claims 1-8 wherein and the guided occupant and the robot when it is detected that has a non-contact state, to stop the movement by the legs The robot according to any one of the above. The robot according to claim 1 , wherein the movement path is determined in consideration of a sum of widths of the robot and the guided person. A control method for a robot having an arm portion having a hand, a leg portion, and a torso portion, and guiding and moving the guided person by pulling the guided person with the hand,
Generating gait data according to the movement route from the movement start point to the movement end point;
Calculating a target position of the hand at a predetermined robot position based on the gait data;
Estimating a follow-up state of the guided person with respect to the robot according to a hand position deviation that is a deviation between the target position of the hand and the measurement position of the hand at the predetermined robot position;
Adjusting the movement speed of the robot according to the estimation result of the following state;
Driving the legs to move the robot at the adjusted moving speed, and
A robot control method for calculating a target position of the hand so that a lateral fluctuation amount of the target position of the hand with respect to the movement path is smaller than a lateral fluctuation quantity of the body portion position with respect to the movement path. The robot control method according to claim 11 , further comprising an impedance control step of impedance-controlling the position of the hand according to a sensor signal from a sensor unit provided in the hand and a target position of the hand. The impedance control step includes controlling the impedance so that the guided person is drawn in a traveling direction along the movement path and the traveling direction is softer than a direction orthogonal to the traveling direction. Item 13. A robot control method according to Item 12 . The robot control method according to any one of claims 11 to 13 , wherein a moving speed of the robot is changed by changing a stride length of the leg portion in accordance with the hand position deviation. The robot control method according to claim 14 , wherein a footed position after changing the stride is calculated along the movement path. 16. The robot control method according to claim 14 or 15 , wherein a footing position after changing the stride is determined on a straight line passing through a virtual footing position when the free leg lands at a reference stride. The robot control method according to claim 16 , wherein the stride changes linearly according to the hand position deviation. The stride length is adjusted according to any one of claims 14 to 17 , wherein when the hand position deviation is smaller than a threshold value, the stride position is adjusted so that a footed position after changing the stride is aligned with a legged position of a support leg. The robot control method described. By a sensor signal from the sensor unit provided in the hand, claim wherein and the guided occupant and the robot when it is detected that has a non-contact state, stopping the movement by the legs 11 to 18 The robot control method according to any one of the above. The robot control method according to any one of claims 11 to 19 , wherein the movement path is determined in consideration of a sum of widths of the robot and the guided person.

Images