JP2008126333A - Biped mobile robot, and method of its walking plan - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a biped mobile robot which can land a sole to an appropriate position when the biped mobile robot walks on a road surface having steps, and further to provide a method of making its walking plan. <P>SOLUTION: First, a candidate position on a floor surface to be landed by the bottom surface of a sole 131 is determined. Next, the central portion of the floor surface facing the central portion Fc of the grounding region of the bottom surface of the sole 131 is obtained when the bottom surface of the sole 131 is landed on the determined candidate position. Then, it is judged whether or not a recessed step with respect to a reference plane including the central portion of the floor surface exists within the region on the floor surface corresponding to the grounding region. When the existence of the recessed step has been judged, the candidate position to land the bottom surface of the sole 131 is corrected such that the area occupied by the recessed step within the corresponding region on the floor surface becomes smaller than a required area. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a walking planning method for a robot having a plurality of leg links connected to a trunk and capable of moving by a walking motion using the plurality of leg links (hereinafter referred to as a legged mobile robot). In particular, the present invention relates to a method for determining a landing position according to the level difference of a walking surface on which a legged mobile robot moves.

An apparatus and a method for making a walking plan for a legged mobile robot have been proposed. For example, in Patent Document 1, a plane group consisting of a reference plane on which a robot is located and a plane substantially parallel to the reference plane is detected from an input image, and an obstacle present on each plane constituting the plane group. An environment recognizing apparatus is disclosed that creates an environment map group having obstacle information for each plane constituting the detected plane group by recognizing. Further, Patent Document 1 discloses a path planning apparatus that determines a movement path for a robot to move from a current position to a target position with reference to an environment map group obtained by the environment recognition apparatus. Specifically, when there is an obstacle on the reference plane on which the robot is located, the route planning apparatus does not consider only the movement path that avoids the obstacle on the reference plane, but the reference plane. In addition to route candidates, a route plan is also made for a movement route using another plane parallel to the plane (for example, a plane including the upper surface of an obstacle existing in the reference plane). As a result, for example, it is possible to plan a movement route that reaches the target position via planes having different heights by going up and down stairs or the like.
JP 2005-92820 A

  As described above, a step existing on the walking surface on which the legged mobile robot walks is detected based on input images such as a stereo image and a distance image, and a moving path involving the detected step is determined. 2. Description of the Related Art Conventionally, a walking planning device and a walking planning method exist. However, in these prior arts, a method for determining an appropriate landing position of the leg link of the legged mobile robot is insufficient, and the legged mobile robot is inherently movable when moving a walking surface on which a step exists. There is a risk of determining that the plane is not movable.

  In Patent Document 1 described above, when it is determined whether or not it is possible to move from the current plane where the robot is located to another plane having a height different from the current plane, the other plane can move the robot. If it is not so wide, it is described that it should not be determined that it can move to another plane. However, there is no specific disclosure regarding what conditions the detected plane area is specifically determined to be able to move. For example, if only a plane wider than the area of the foot provided at the tip of the leg link is determined as a movable plane, there is a high possibility that a safe movement path can be determined. However, since it is assumed that stable landing is possible even if there is a step under the foot, it is necessary to use an appropriate route so that such a highly safe moving route can quickly reach the target position. Not always.

  The present invention has been made in consideration of the above-described circumstances, and when moving a walking surface on which a step exists in the legged mobile robot, the foot provided at the tip of the leg link of the legged mobile robot is appropriate. It is an object of the present invention to provide a walking planning method and a legged mobile robot that can land at various positions.

  The walking planning method according to the first aspect of the present invention includes a plurality of legs, and a foot is provided at the lower end of each of the legs via an ankle joint. This is a walking planning method for a legged mobile robot that walks on the floor surface by repeating the operation of grounding the bottom surface on the floor surface. In this method, first, candidate positions for grounding the bottom surface are determined on the floor surface. Next, when landing the bottom surface based on the candidate position, a corresponding center portion on the floor surface facing the center portion of the ground contact area of the bottom surface is obtained. Subsequently, it is determined whether or not there is a concave step with respect to a reference plane including the corresponding central portion in the corresponding region on the floor surface corresponding to the ground contact region. And when it determines with the said concave step existing, the said candidate position is corrected so that the area which the said concave step occupies in the said corresponding area may be below a predetermined area.

  By such a method, when the presence of a concave step is detected at a candidate position where the foot is to be landed, the landing position can be determined by allowing the foot to protrude above the concave step. it can. Thereby, the area | region which can be selected as a landing position of a foot can be expanded compared with the strict method which selects only the area | region larger than the area of the whole contact | abutting area | region of a foot bottom as the area | region where a foot can land. That is, it is possible to appropriately determine the foot landing position.

  The walking planning method according to the first aspect further determines whether or not a convex step exists in the corresponding region with respect to the plane including the corresponding central portion, and determines that the convex step exists. In this case, it is desirable to avoid the convex step and to correct the candidate position at a position where a predetermined gap is provided between the convex step and the foot portion. As described above, since a predetermined gap is provided between the convex step and the foot is landed, an interval is generated between the foot and the convex step when the leg is lifted when the legged mobile robot is walking. It is possible to prevent the foot from being caught on the convex step.

  Further, in the walking planning method according to the first aspect, the presence determination of the convex step or the concave step is referred to environmental map data created by recognizing a plurality of planes existing on the floor surface. It is desirable to perform the determination based on whether or not a plurality of planes having different line vectors exist in the corresponding region. By determining whether or not there are different planes in this way, the step search speed can be improved. That is, if there is no different plane, it can be quickly determined that there is no step.

The walking planning method according to the second aspect of the present invention comprises a plurality of legs, each of which has a foot part provided through an ankle joint at the lower end of each of the leg parts. This is a walking planning method for a legged mobile robot that walks on the floor surface by repeating the operation of grounding the bottom surface on the floor surface, and this method performs the following processes (1) to (9): Features.
(1) The step length candidate value is set as the initial step length.
(2) A candidate position for landing the bottom surface is calculated using the step candidate value.
(3) When the bottom surface is landed based on the candidate position, a corresponding center portion on the floor surface facing the center portion of the ground contact area of the bottom surface is obtained.
(4) It is determined whether or not there is a convex step or a concave step with respect to a reference plane including the corresponding central portion in the corresponding region on the floor corresponding to the ground contact region.
(5) If it is determined that the convex step is present in front of the corresponding central portion in the corresponding region, the step length to a position where a gap between the boundary of the convex step and the foot portion is at least a first value. Reduce the candidate value.
(6) When it is determined that the concave step exists in front of the corresponding central portion in the corresponding region, the protruding amount of the bottom surface from the boundary of the concave step to the front side is a second value or less. The step candidate value is reduced.
(7) When it is determined that the convex step is present behind the corresponding central portion in the corresponding region, the protrusion amount of the bottom surface from the boundary of the convex step to the front side is a third value or less. The step candidate value is reduced.
(8) When it is determined that the concave step exists behind the corresponding central portion in the corresponding region, the protruding amount of the bottom surface from the boundary of the concave step to the front side is a value equal to or less than a fourth value. The step candidate value is reduced.
(9) The landing position of the foot is determined based on the step candidate value obtained by repeating the processes (2) to (8) until no convex step or concave step is detected in the corresponding region. To do.

  In other words, in the method according to the second aspect, a step existing on the walking surface is detected, the detected step type is classified into four patterns, and the foot landing position is determined according to the detected step classification result. Was decided to be corrected. By adopting such an algorithm, the foot landing position determination process can be systematically performed, and the foot landing position can be quickly planned.

  A legged mobile robot according to a third aspect of the present invention is a legged mobile robot that walks on a floor surface, and includes a trunk, a plurality of legs connected to the trunk, and the plurality of legs. A three-dimensional map of the floor surface based on a plurality of foot portions provided at the lower ends of the respective portions via an ankle joint, a visual sensor for visually recognizing the walking surface, and information acquired by the visual sensor A map creation unit that creates data; and a walking plan unit that determines landing positions of the plurality of foot portions using the three-dimensional map data. Further, the walking planning unit determines a candidate position for landing the bottom surface of the foot portion on the floor surface, and when the bottom surface is landed based on the candidate position, a center of the ground contact area of the bottom surface A corresponding center portion on the floor surface opposite to the portion is obtained, and whether or not there is a concave step with respect to a reference plane including the corresponding center portion in the corresponding region on the floor surface corresponding to the ground contact region. When the determination is made with reference to the three-dimensional map data and it is determined that the concave step exists, the candidate position is corrected so that the area occupied by the concave step in the corresponding region is equal to or less than a predetermined area. It is characterized by that.

  With such a configuration, when the presence of a concave step is detected at a candidate position where the foot is to be landed, the landing position can be determined by allowing the foot to protrude above the concave step. it can. Thereby, the area | region which can be selected as a landing position of a foot can be expanded compared with the strict method which selects only the area | region larger than the area of the whole contact | abutting area | region of a foot bottom as the area | region where a foot can land. That is, it is possible to appropriately determine the foot landing position.

  According to the present invention, it is possible to provide a walking planning method and a legged mobile robot capable of landing a foot at an appropriate position when a legged mobile robot moves a walking surface where a step exists.

  Hereinafter, specific embodiments to which the present invention is applied will be described in detail with reference to the drawings. In the drawings, the same elements are denoted by the same reference numerals, and redundant description will be omitted as necessary for the sake of clarity.

Embodiment 1 of the Invention
The robot 100 according to the present embodiment is a legged mobile robot having two leg links. Further, the robot 100 has a walking planning device 1. The walking planning device 1 is a device that determines the landing position of the foot provided at the tip of each of the two leg links of the robot 100 with reference to environmental map data including three-dimensional information of the walking surface on which the robot walks. is there. The robot 100 generates motion data that realizes the foot landing position determined by the walking planning device 1, and walks by driving a joint group of the robot 100 with an actuator based on the generated motion data.

  First, the joint degrees of freedom of the robot 100 will be described with reference to FIG. FIG. 1 is a model diagram showing the robot 100 by a link connecting the joints. The robot 100 includes a head 101, two leg links LR and LL, two arm links AR and AL, and a trunk BD to which these are connected.

  The head 101 of the robot 100 is provided with a visual sensor 101 that acquires external three-dimensional point group data, that is, distance image data. The neck joint that supports the head 101 includes a joint 102 in the roll direction, a joint 103 in the pitch direction, and a joint 103 in the yaw direction. The right arm link AR has a joint 105 in the shoulder pitch direction, a joint 106 in the shoulder roll direction, a joint 107 in the upper arm yaw direction, a joint 108 in the pitch direction of the elbow, and a joint 109 in the yaw direction of the wrist. A hand 141R is provided at the end of the AR. Note that the mechanism of the hand portion 141R may be determined depending on the shape, type, and the like of the object to be held. For example, it may be a multi-joint structure having a plurality of fingers and a structure having multiple degrees of freedom.

  The left arm link AL has the same structure as the right arm link AR. Specifically, the left arm link AL has five joints 110 to 114, and has a hand portion 141L at the end thereof.

  The right leg link LR includes a hip yaw joint 118, a hip pitch joint 119, a waist roll joint 120, a knee pitch joint 121, an ankle pitch joint 122, and an ankle roll direction. Joint 123. A foot 131 </ b> R that is in contact with the walking surface is provided below the ankle joint 123.

  The left leg link LL has the same structure as the right leg link LR. Specifically, the left leg link LL has six joints 124 to 129, and has a foot 131L at the end thereof.

  The trunk 143 includes a joint 115 in the yaw direction, a joint 116 in the roll direction, and a joint 117 in the pitch direction.

  Subsequently, a control system for walking the robot 100 will be described below. The configuration of the control system of the robot 100 is shown in FIG. In FIG. 2, the visual sensor 10 acquires the distance image data of the outside world of the robot 100 as described above. Specifically, distance image data is acquired by an active distance sensor such as a laser range finder. It has a plurality of cameras equipped with an image sensor such as a CCD (Charge Coupled Device) image sensor or a CMOS (Complementary Metal Oxide Semiconductor) image sensor, and generates distance image data using image data captured by these cameras. May be. Specifically, corresponding points are detected from image data captured by a plurality of cameras, and the three-dimensional positions of the corresponding points are restored by stereo viewing. Here, the search for corresponding points in a plurality of captured images may be performed by applying a known method such as a gradient method or a correlation method using a time-space differential constraint formula for the plurality of captured images.

  The walking plan apparatus 1 includes an environment map generation unit 11 and a walking plan unit 12. The environment map generation unit 11 detects a plane from the distance image data and generates an environment map as a set of a plurality of planes. More specifically, the environment map of the present embodiment is generated as grid data obtained by dividing a two-dimensional plane (referred to as an xy plane) into a grid. As data corresponding to each grid of the environmental map, the height in the z-axis direction perpendicular to the xy plane of each grid, the plane ID that can uniquely identify the plane to which each grid belongs, and the normal of the plane to which each grid belongs Vector is retained. The environment map generation unit 11 recognizes a set of adjacent grids having substantially the same height as one plane, and assigns a unique plane ID to the recognized plane.

  For example, when the robot 100 walks on the walking surface 160 illustrated in FIG. 3A, the planes P1 to P8 are detected by the environment map generation unit 11, and the environment map 20 as illustrated in FIG. 3B is generated. . As the data of each grid of the environmental map 20, the plane ID that can uniquely identify each of the planes P1 to P8, the height of each grid in the z-axis direction, and the normal vector (na, nb, nc) are held. . As will be described later, in order to detect road surface unevenness at the landing positions of the feet 131R and L based on the difference in the plane ID between adjacent grids and the difference in the height in the z-axis direction, The area of the grid may be selected to be sufficiently smaller than the areas of the bottom surfaces of the feet 131R and L.

  When generating a wide-range environmental map, the distance image data after acquiring the distance image data at a plurality of different viewpoints by moving the visual sensor 10, that is, the head 101, and integrating the distance image data at the plurality of viewpoints. Generate an environment map based on Here, the integration of the distance image data obtained from a plurality of viewpoints is performed by aligning the plurality of distance image data based on the angle of the neck joint when the distance image data is acquired by the visual sensor 10. Alternatively, the distance image data may be aligned with each other by obtaining corresponding points from a plurality of distance image data.

  The walking plan unit 12 determines a target position with reference to the environment map created by the environment map generation unit 11, and calculates the landing positions of the feet 131R and L for reaching the determined target position. In addition, the detail of the determination method of the foot landing position by the walk plan part 12 is mentioned later.

  The motion generation unit 13 generates motion data for realizing the landing positions of the feet 131R and L generated by the walking planning device 1. Here, the motion data includes the ZMP position, the center of gravity position of the robot 100, the positions and postures of the feet 131R and L, and the position and posture of the trunk BD.

  The control unit 14 receives the motion data generated by the motion generation unit 13 and calculates a target joint angle of each joint by inverse kinematics calculation. Further, the control unit 14 calculates a torque control value for driving each joint based on the calculated target joint angle and the current joint angle measured by the encoder 16. The robot 100 is walked by operating the actuator 15 for driving each joint according to the torque control value calculated by the control unit 14.

  Each time the robot 100 advances one step, the determination of the landing position of the foot 131 by the walking planning device 1 may be repeated, or the landing positions of the feet 131R and L may be determined in units of a plurality of steps. The walking motion may be performed based on the above.

  Subsequently, the procedure for determining the landing positions of the feet 131R and L in the walking plan unit 12 will be described below with reference to the flowcharts shown in FIGS. First, in step S101, the stride W when stepping on the foot 131 is set to the reference stride Wr. In the following description, when it is simply indicated as the foot 1130, one of the two feet 131R and L, specifically, the foot of the free leg to be determined as the landing position is indicated. To do.

  In step S102, when it is assumed that the foot 131 of the free leg has been stepped on with the stride W, the unevenness of the walking surface is searched for in the foot enlarged region Ra. Here, the foot enlargement region Ra is a region obtained by combining the ground contact region of the foot 131 and the region enlarged by the width La in the front-rear direction of the foot 131, as shown in FIG. .

  A specific walking surface unevenness search can be performed, for example, as follows. First, by referring to the environment map, it is determined whether or not there are two planes having different plane IDs on the walking plane in contact with the foot enlargement area Ra. Thus, by performing the comparison of the plane IDs first, the unevenness search speed can be improved. That is, if the plane ID is the same, it can be quickly determined that there is no unevenness. In addition, it is possible to quickly determine an area where there is a possibility of unevenness due to the difference in the plane ID.

  When there are two different planes on the walking surface in contact with the foot enlargement area Ra, the height zi of the foot 131 when assuming that the foot 131 is landed on the plane in contact with the foot center Fc, The difference between the height zp of the other plane different from the plane in contact with the foot center Fc is compared with a predetermined threshold value Hs. If the absolute value of the difference between the height zi and the height zp at this time is larger than the threshold value Hs, it is determined that there is a plane that is a convex step or a concave step when viewed from the plane that the foot center Fc contacts. Here, the foot center Fc means the center position of the ankle joints 122 and 123 or 128 and 129. That is, if zp-zi is greater than the threshold value Hs, it can be determined that there is a plane that has a convex step when viewed from the foot center Fc. On the other hand, if zi-zp is larger than the threshold value Hs, it can be determined that there is a plane that is a concave step when viewed from the foot center Fc. In the following, a plane that is a convex step when viewed from the foot center Fc is simply referred to as a “convex step”, and a plane that is a concave step when viewed from the foot center Fc is simply referred to as a “concave step”.

  For example, in the walking surface 160 of FIG. 2, the example of FIG. 7A in which the foot center Fc is located on the plane P7 and the example of FIG. 7B in which the foot center Fc is located on the plane P4. Then, when zi-zp is larger than the threshold value Hs on the front side of the foot center Fc, the presence of a concave step is detected on the front side of the foot center Fc.

  When it is detected in the above-described search in step S102 that there is a convex step in the foot enlargement area Ra, the processes in steps S104 to S109 are executed (step S103). In step S104, it is determined whether or not the front end of the plane detected as the convex step is located behind the foot center Fc. When the determination in step S104 is established, it is determined that the detected convex step is a step corresponding to the pattern C among the four types of patterns of steps A to D (step S105).

As shown in FIG. 8, the step divided into the pattern C is a form in which a convex step exists behind the foot center Fc. Landing the foot 131 in a place where there is a convex step divided into the pattern C is likely to cause unstable behavior such as the robot 100 falling. For this reason, the stride W is reduced so that the front end of the foot 131 is located at a position where the protrusion amount Lb is left behind from the front end of the plane detected as the convex step (step S105). The reduction width S of the stride W at this time is as shown in FIG.
S = f1-Lb + d
Can be calculated. Here, f1 is the distance between the foot center Fc and the front end of the foot 131. The distance d is a distance between the foot center Fc and the front end of the convex step. Assuming a step perpendicular to the x-axis as shown in FIG. 3A, the distance d can be calculated by the difference between the x coordinate xc of the foot center Fc and the x coordinate xmax of the front end of the convex step.

  On the other hand, when it is determined that both the front end and the rear end of the plane detected as the convex step are located in front of the foot center Fc, the detected convex step is the presence pattern of the four types of steps A to D. Is determined to be a step corresponding to the pattern A (steps S106 and S107).

As shown in FIG. 8, the step divided into the pattern A is a form in which a convex step exists ahead of the foot center Fc. When the foot 131 is landed at the place where the convex step classified into the pattern A exists, the robot 100 falls down due to the presence of the convex step at the landing place of the foot 131, or the front end of the foot 131 and the convex step Therefore, there is a high possibility that the foot 131 will trip over a convex step when kicking the foot 131 again as a free leg. For this reason, the stride W is reduced so that the front end of the foot 131 is positioned at the position where the rear end of the plane detected as the convex step and the gap La is generated (step S107). The reduction width S of the stride W at this time is as shown in FIG.
S = f1 + La-d
Can be calculated. Here, f1 is the distance between the foot center Fc and the front end of the foot 131. The distance d is a distance between the rear end of the convex step and the foot center Fc. Assuming a step perpendicular to the x-axis as shown in FIG. 3A, the distance d can be calculated by the difference between the x-coordinate xmin of the rear end of the convex step and the x-coordinate xc of the foot center Fc.

  When the detected convex step is not classified into any of the patterns A and C described above, the position where the rear end of the convex step and the gap La are generated in order to avoid landing on such a convex step. The stride W is reduced so that the front end of the foot 131 is positioned at step S108.

  In step S109, it is determined whether the reduced stride W is equal to or greater than zero. If the stride W is equal to or greater than zero, the process returns to step S102 and the process is repeated. On the other hand, if the stride W is smaller than zero, the process branches to step S117 shown in FIG. 5 to set the stride W to zero, and outputs this as a confirmed stride (step S118).

  Next, the processing contents of steps S110 to S116 that are executed when no convex step is detected in the foot enlargement area Ra in step S103 described above will be described with reference to FIG. In step S110, it is determined whether or not there is a concave step in the foot reduction region Rb based on the result of the search in step S102 described above. Here, the foot reduction region Rb is a partial region of the ground expansion region Ra and the ground contact region of the foot 131. As shown in FIG. This is a region reduced from the front end and the rear end of the flat 131 by a width Lb.

  If it is determined in step S110 that there is a concave step in the foot reduction region Rb, the processing in steps S111 to S116 is executed. In step S111, it is determined whether or not the front end of the plane detected as the concave step is located behind the foot center Fc. If the determination in step S111 is true, it is determined that the detected concave step is a step corresponding to the pattern D among the four types of stepped patterns of patterns A to D (step S112).

As shown in FIG. 8, the step divided into the pattern D is a form in which there is a concave step behind the foot center Fc in the foot reduction region Rb. Landing the foot 131 at a place where there is a concave step divided into the pattern D is likely to cause unstable behavior such as the robot 100 falling over by stepping off the step when the free leg lands. For this reason, the stride W is reduced so that the front end of the foot 131 is located at a position where the protrusion amount Lb is left behind from the front end of the plane detected as the concave step (step S112). The reduction width S of the stride W at this time is as shown in FIG.
S = f1-Lb + d
Can be calculated. Here, f1 is the distance between the foot center Fc and the front end of the foot 131. The distance d is the distance between the foot center Fc and the front end of the concave step. Assuming a step perpendicular to the x-axis as shown in FIG. 3A, the distance d can be calculated by the difference between the x coordinate xc of the foot center Fc and the x coordinate xmax of the front end of the concave step.

  On the other hand, when it is determined that both the front end and the rear end of the plane detected as the concave step are located in front of the foot center Fc, the detected concave step is the presence pattern of the four types of steps of patterns A to D. Is determined to be a step corresponding to the pattern B (steps S113 and S114).

As shown in FIG. 8, the step divided into the pattern B is a form in which there is a concave step in front of the foot center Fc in the foot reduction region Rb. Landing the foot 131 in a place where there is a concave step divided into the pattern B is likely to cause unstable behavior such as the robot 100 falling down by stepping off the step when the free leg lands. For this reason, the stride W is reduced so that the front end of the foot 131 is located at the position where the protrusion amount Lb is left behind from the rear end of the plane detected as the concave step (step S114). The reduced width S of the stride W at this time is as shown in FIG.
S = f1-Lb-d
Can be calculated. Here, f1 is the distance between the foot center Fc and the front end of the foot 131. The distance d is a distance between the rear end of the concave step and the foot center Fc. Assuming a step perpendicular to the x-axis as shown in FIG. 3A, the distance d can be calculated by the difference between the x-coordinate xmin of the rear end of the concave step and the x-coordinate xc of the foot center Fc.

  When the detected concave step is not classified into any of the patterns B and D described above, a gap Lb is left from the rear end of the concave step to avoid landing on the concave step. The stride W is reduced so that the front end of the foot 131 is positioned at the position (step S115).

  In step S116, it is determined whether the reduced stride W is equal to or greater than zero. If the stride W is equal to or greater than zero, the process returns to step S102 and the process is repeated. On the other hand, if the stride W is smaller than zero, the process branches to step S117, the stride W is set to zero, and this is output as a confirmed stride (step S118).

  As a result of repeating the step correction processing according to the detection of the convex step and the concave step described above, when neither the convex step in the foot enlargement region Ra nor the concave step in the foot contraction region Rb is detected, The stride at this time is output as a confirmed stride (steps S110 and S118).

  As described above, it is possible to cope with steps of various shapes existing on the walking surface on which the robot 100 moves by repeating the correction of the stride while classifying the detected steps into any of the patterns A to D. Can do. For example, when the assumed landing position of the foot 131 is the boundary position between the planes P3 and P4 where the change in the slope exists as shown in the upper part of FIG. 10, it is determined as the convex step of the pattern C in the first processing loop. As a result, the stride length is reduced with the protrusion amount Lb from the rear end of the plane P4. Next, the landing position after the reduction of the stride is determined as a convex step of the pattern A in the next processing loop as shown in the middle stage of FIG. For this reason, as shown in the lower part of FIG. 10, the stride is reduced to the position where the front end of the plane P3 and the gap La are generated.

  Further, for example, when the assumed landing position of the foot 131 is the boundary position between the planes P1 and P2 where the stepped step as shown in the upper part of FIG. 11 exists, the concave portion of the pattern D is formed in the first processing loop. The step is determined to be a step, and the stride is reduced by leaving the amount of protrusion Lb from the rear end of the plane P2. Next, the landing position after the reduction of the stride is determined as a convex step of the pattern A in the next processing loop as shown in the middle part of FIG. For this reason, as shown in the lower part of FIG. 11, the stride is reduced to the position where the front end of the plane P1 and the gap La are generated.

  As described above, when the presence of a concave step is detected at a candidate position where the foot 131 is to be landed, the walking planning device 1 allows the foot 131 to protrude beyond the concave step. Determine the landing position. Thereby, the area | region which can be selected as a landing position of the foot 131 can be expanded compared with the strict method which selects only the area | region larger than the area of the whole bottom face of the foot 131 as the area | region where a foot can land. That is, it is possible to appropriately determine the foot landing position.

  Furthermore, when the presence of a convex step is detected at the candidate position where the foot 131 is to land, the walking plan device 1 provides a predetermined gap La between the convex step and the foot 131. Determine the landing position. As described above, since the foot 131 is landed with a predetermined gap between the convex step and the foot 131 is lifted when the robot 100 is walking, an interval is generated between the foot 131 and the convex step. Further, it is possible to prevent the foot 131 from being caught on the convex step.

  Furthermore, the walking planning device 1 detects the level difference existing on the walking surface, classifies the detected level types into the four patterns A to D described above, and sets the step according to the detected level classification result. I decided to adjust the landing position of the flat 131. By adopting such an algorithm, the determination process of the landing position of the foot 131 can be systematically performed, so that the planning of the landing position of the foot 131 can be executed promptly.

Embodiment 2 of the Invention
In Embodiment 1 of the invention, the walking planning device 1 has been described as creating an environment map as one two-dimensional grid data. In this case, it is necessary to calculate the height zi of the foot 131 required in step S102 shown in FIG. 4 every time the step search of the foot enlargement region Ra is performed. The height zi of an arbitrary point (xi, yi, zi) on the bottom surface of the foot 131 is determined by the plane normal vector (na, nb, nc) and the position (xc, This can be obtained by substituting the xy coordinates (xi, yi) of the search point to be calculated into a plane equation determined by yc, zc).

  The walking planning device 2 according to the present embodiment includes an environmental map generating unit 21 instead of the environmental map generating unit 11 included in the walking planning device 1, and includes a walking planning unit 22 instead of the walking plan unit 12. The environment map generation unit 21 detects a plane in the same manner as the environment map generation unit 11, generates 2D grid data for each detected plane, and generates an environment map as a set of a plurality of 2D grid data. At this time, the “real height” or “temporary height” in the z-axis direction is held in the grid data corresponding to one plane corresponding to each grid. Here, the “real height” is the actual height of the walking surface, and is a value held in the grid of the region where the walking surface actually exists. On the other hand, the “temporary height” is a value held in a grid in a region where the walking surface does not actually exist, and extends the detected plane to a region where the walking surface does not actually exist. This is the height of the plane calculated when it is assumed.

  For example, when recognizing the walking surface 260 shown in FIG. 12, the environment map generation unit 21 detects three planes P <b> 11 to P <b> 13. Further, the environment map generation unit 21 generates an environment map 30 including three two-dimensional grid data 31 to 32 corresponding to the detected three planes. Grid data 31 corresponding to the plane P11 includes a grid group 311 that holds the actual height (a grid indicated by a blank in FIG. 13) and a grid group 312 that holds a temporary height (a grid indicated by diagonal lines in FIG. 13). )including. Similarly, the grid data 32 corresponding to the plane P12 includes a grid group 321 that holds the actual height and a grid group 322 that holds the temporary height. Similarly, the grid data 33 corresponding to the plane P13 includes a grid group 331 that holds the actual height and a grid group 332 that holds the temporary height.

  The walking plan unit 22 plans the landing position of the foot 131 similar to the walking plan unit 11 of the first embodiment of the invention, but determines the height zi of the foot 131 necessary in step S102 shown in FIG. This is obtained by reading out the “real height” or “temporary height” held as an environmental map.

  In the present embodiment, since the environment map generation unit 21 generates two-dimensional grid data for each detected plane, the memory required to hold the environment map compared to the first embodiment of the invention. The amount increases. However, the walking plan unit 22 does not need to calculate the height zi of the foot 131 every time the search point is changed when searching for a step existing in the foot enlarged region Ra. It is possible to improve the processing speed when planning the landing position.

Other embodiments.
In the first embodiment of the present invention described above, the processing of the walking planning apparatus 1 that is adapted to a case where a step boundary exists along the left-right direction of the foot 131 has been described for the sake of simplicity. However, it is easy to extend the processing of the above-described walking planning device 1 in the case where there is a step boundary along the front-rear direction of the foot 131.

  Specifically, the foot enlargement area Ra may be enlarged also in the left-right direction of the foot 131. Further, the foot reduction region Rb may be reduced in the left-right direction of the foot 131 as compared with the ground contact region of the foot 131. Further, the walking plan unit 22 allows the foot 131 to protrude on the concave step by a predetermined protrusion amount Lb ′ when there is a boundary line of the concave step along the front-rear direction of the foot 131. What is necessary is just to determine a position. Furthermore, the walking plan unit 22 lands so as to land with a predetermined gap La ′ between the convex step and the foot 131 when there is a boundary line of the convex step along the front-rear direction of the foot 131. The position may be determined.

  That is, when the walking planning device 1 assumes that the foot 131 is landed at the candidate position on the walking surface, the walking surface facing the bottom surface of the foot 131 with the plane facing the foot center Fc as a reference surface. When there is a region having a concave step that is lower than the reference surface in the upper corresponding region, the foot 131 is set so that the area occupied by the concave step in the corresponding region on the walking surface is equal to or less than a predetermined area. It is better to determine the position where the landing is made. Here, the predetermined area may be appropriately set within a range in which the robot 100 can be supported. For example, in Embodiment 1 of the invention, a foot reduction region Rb smaller than the area of the bottom surface of the foot is defined, and a concave step is formed in a partial region corresponding to the foot reduction region Rb among the corresponding regions on the walking surface. The landing position of the foot 131 is determined so as not to be included.

  In addition, when the walking planning device 1 assumes that the foot 131 is to be landed at a candidate position on the walking surface, the walking surface facing the bottom surface of the foot 131 with a plane facing the foot center Fc as a reference surface. A position having a predetermined gap between the protruding step and the foot 131 when the protruding step having a height higher than the reference surface is present in the corresponding region above. In addition, the landing position of the foot 131 may be corrected. By such an extension, the landing position of the foot 131 can be appropriately determined regardless of the angle of the convex step or the concave step with respect to the traveling direction of the foot 131.

  In Embodiment 1 of the invention, both the convex and concave steps of the walking surface are detected and the landing position of the foot 131 is corrected so as to avoid them. It is also possible to configure to perform only one of the detection and avoidance operations. By executing at least the detection and avoidance process of the concave step, the area that can be selected as the landing position of the foot 131 can be enlarged, and the foot landing position can be appropriately determined.

  In the first embodiment of the invention, the gap La and the protruding amount Lb are common values on the front side and the rear side of the foot 131, but these values may be different.

  In the first embodiment of the invention, the case where the present invention is applied to the legged mobile robot 100 that performs autonomous movement has been described. However, for example, the present invention can be applied to a legged mobile robot that moves according to the operation of the passenger. In this case, if the walking plan unit 12 is configured to set the target position in the direction input by the operation of the control stick or the like by the passenger and sequentially determine the landing position of the foot 131 toward the target position. Good.

  In addition, the walking planning apparatus 1 according to the first embodiment of the present invention initially sets the maximum stride Wr as the initial stride, and sequentially reduces the stride when a convex step or a concave step is detected. . However, the initial stride may be set to a value smaller than the maximum stride of the robot 100, and when a convex step or a concave step is detected, the step may be avoided by widening the stride. For example, when the initial step length is set to a value smaller than the maximum step length by the length obtained by adding the gap La to the entire length of the foot 131 and the above-described concave step of the pattern B or the convex step of the pattern C is detected, A position where the stride is widened forward by the gap La from the boundary can be set as the landing position of the foot.

  Further, the feet 131R and L included in the robot 100 according to the first embodiment of the invention have been described as having no joints. However, for example, the toes may be provided at the tips of the feet 131R and L via joints. In the legged mobile robot configured as described above, when the foot link that is the supporting leg becomes a free leg and the foot leaves the floor surface, only the toe portion may come into contact with the ground. For this reason, what is necessary is just to set so that the protrusion amount Lb may become smaller than the length of a toe part so that a concave level | step difference may not be stepped off also in the attitude | position which only a toe part grounds. Alternatively, when the protrusion of the foot 131 on the concave step is permitted, the operation data may be created by prohibiting the posture in which only the toe portion contacts the ground.

  Furthermore, the present invention is not limited to the above-described embodiment, and various modifications can be made without departing from the gist of the present invention described above.

It is a model diagram of a legged mobile robot according to the present invention. It is a block block diagram of the walking plan apparatus concerning this invention. It is a figure which shows an example of the walking surface to which the legged mobile robot concerning this invention moves, and an example of an environmental map. It is a flowchart which shows the processing content of the walking plan apparatus concerning this invention. It is a flowchart which shows the processing content of the walking plan apparatus concerning this invention. It is a figure for demonstrating a foot expansion area | region and a foot reduction area | region. It is a figure for demonstrating the level | step difference detection process which the walk planning apparatus concerning this invention performs. It is a figure for demonstrating the level | step difference detection process which the walk planning apparatus concerning this invention performs. It is a figure for demonstrating the level | step difference detection process which the walk planning apparatus concerning this invention performs. It is a figure for demonstrating the level | step difference detection process which the walk planning apparatus concerning this invention performs. It is a figure for demonstrating the level | step difference detection process which the walk planning apparatus concerning this invention performs. It is a figure which shows an example of the walk surface to which the leg type mobile robot concerning this invention moves. It is a figure which shows an example of the environmental map produced | generated by the leg type mobile robot concerning this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1, 2 Walking plan apparatus 10 Visual sensor 11, 21 Environmental map production | generation part 12, 22 Walk planning part 13 Operation | movement production | generation part 14 Control part 15 Actuator 16 Encoder 20, 30 Environmental map 100 Legged mobile robot 101 Head 102-129 Joint 131R, 131L Foot 141R, 141L Hand 160, 260 Walking surface LR Right leg link LL Left leg link AR Right arm link AL Left arm link BD Trunk

Claims (13)

A plurality of legs are provided, and a foot portion is provided at each lower end of each of the plurality of leg portions via an ankle joint, and the operation of grounding the bottom surface of the foot portion on the floor surface is repeated. A walking planning method for a legged mobile robot walking on a surface,
On the floor surface, determine a candidate position for grounding the bottom surface,
When landing the bottom surface based on the candidate position, obtain a corresponding center portion on the floor surface facing the center portion of the ground contact area of the bottom surface,
In the corresponding region on the floor corresponding to the grounding region, it is determined whether there is a concave step with respect to a reference plane including the corresponding central portion,
When it is determined that the concave step exists, the candidate position is corrected so that the area occupied by the concave step in the corresponding region is equal to or less than a predetermined area.
A walking planning method characterized by that.   In the correction of the candidate position, the candidate position is determined so that the concave step is not included in the corresponding partial area in the corresponding area corresponding to the partial area including the central portion of the ground contact area on the bottom surface. The walking planning method according to claim 1, wherein the walking planning method is performed.   The presence of the concave step is determined based on whether or not the difference between the height of the grounding area and the height of the corresponding area is greater than a predetermined threshold when the bottom surface is grounded to the reference plane. Item 2. The walking planning method according to item 1. In the corresponding region, further determine whether or not there is a convex step with respect to the plane including the corresponding central portion,
The said candidate position is corrected to the position which provided the predetermined clearance gap between the said convex step and the said foot part, when it determines with the said convex step existing. Gait planning method.   The presence of the convex step is determined based on whether or not the height of the corresponding region viewed from the grounding region is larger than a predetermined threshold when it is assumed that the bottom surface is grounded to the reference plane. The gait planning method described.   The walking planning method according to claim 1, wherein the central portion is a position obtained by projecting the position of the ankle joint on the bottom surface.   The presence determination of the convex step or the concave step is referred to environmental map data created by recognizing a plurality of planes existing on the floor surface, and a plurality of planes having different normal vectors are used as the corresponding region. The gait planning method according to claim 1 or 4 performed based on whether or not it exists. The environmental map data is a set of two-dimensional grid data created for each plane existing on the floor surface, and each of the two-dimensional grid data virtually extends a plane existing on the floor surface. Holding the virtual height of the plane in the case,
The gait planning method according to claim 7, wherein presence determination of the convex step or the concave step is performed using the virtual height obtained by referring to the two-dimensional grid data. A plurality of legs are provided, and a foot portion is provided at each lower end of each of the plurality of leg portions via an ankle joint, and the operation of grounding the bottom surface of the foot portion on the floor surface is repeated. A walking planning method for a legged mobile robot walking on a surface,
Set the stride candidate value to the initial stride,
A candidate position for landing the bottom surface is calculated using the stride candidate value,
When landing the bottom surface based on the candidate position, obtain a corresponding center portion on the floor surface facing the center portion of the ground contact area of the bottom surface,
In the corresponding region on the floor corresponding to the ground contact region, it is determined whether there is a convex step or a concave step with respect to a reference plane including the corresponding central portion,
If it is determined that the convex step exists ahead of the corresponding central portion in the corresponding region, the step candidate value is set to a position where a gap between the boundary of the convex step and the foot portion is at least a first value. Shrink,
If it is determined that the concave step exists in front of the corresponding central portion in the corresponding region, the step candidate until a position where the protruding amount of the bottom surface from the boundary of the concave step to the front side is equal to or less than a second value Reduce the value,
If it is determined that the convex step exists behind the corresponding central portion in the corresponding region, the step candidate until the position where the protruding amount of the bottom surface from the boundary of the convex step to the front side is equal to or less than a third value Reduce the value,
If it is determined that the concave step exists behind the corresponding central portion in the corresponding region, the step candidate until a position where the amount of protrusion of the bottom surface from the boundary of the concave step to the front side becomes a fourth value or less Reduce the value,
Step candidate values obtained by repeating the candidate position calculation processing, the convex step and concave step existence determination processing, and the step candidate reduction processing until no convex step and concave step are detected in the corresponding region. A stride planning method for determining a landing position of the foot portion based on the step. A legged mobile robot walking on the floor,
The trunk,
A plurality of legs connected to the trunk;
A plurality of foot portions provided via an ankle joint at the lower end of each of the plurality of leg portions;
A visual sensor for visually recognizing the floor;
A map creation unit that creates three-dimensional map data of the floor surface based on information acquired by the visual sensor;
A walking planning unit that determines landing positions of the plurality of foot portions using the three-dimensional map data;
The walking plan section
On the floor surface, determine a candidate position for landing the bottom surface of the foot,
When landing the bottom surface based on the candidate position, obtain a corresponding center portion on the floor surface facing the center portion of the ground contact area of the bottom surface,
In the corresponding region on the floor corresponding to the ground contact region, it is determined with reference to the three-dimensional map data whether there is a concave step with respect to a reference plane including the corresponding central portion,
When it is determined that the concave step is present, the legged mobile robot is configured to correct the candidate position so that an area occupied by the concave step in the corresponding region is equal to or less than a predetermined area.   When the walking plan section assumes that the bottom surface is grounded to the reference plane, whether the difference in height between the ground contact area and the corresponding area is greater than a predetermined threshold or not. The legged mobile robot according to claim 10, wherein presence is determined.   The gait planning unit further determines whether or not a convex step exists in the corresponding region with respect to the plane including the corresponding central portion, and determines that the convex step exists, The legged mobile robot according to claim 10, wherein the candidate position is corrected to a position where a predetermined gap is provided between the convex step and the foot portion.   When the walking plan unit assumes that the bottom surface is grounded to the reference plane, the presence of the convex step is determined based on whether or not the height of the corresponding region viewed from the grounding region is larger than a predetermined threshold. The legged mobile robot according to claim 12.