Motion planning for biped robot

Abstract

The present description relates to motion planning for a biped robot. According to an example of the present disclosure, a motion planning method for a biped robot may comprise: acquiring a discrete map with terrain information that corresponds to an area to be walked, determining terrain points in the discrete map that indicate that a biped robot is allowed to pass through, and generating a preliminary foothold sequence on the basis of the terrain points; generating a preliminary motion trajectory of the biped robot on the basis of the discrete map; and on the basis of a preset objective function and on the basis of the preliminary motion trajectory and a kinematic model of the biped robot, optimizing the preliminary foothold sequence, so as to generate a target foothold sequence. By means of determining, on the basis of gradient information of the ground level in a discrete map, whether a position allows a biped robot to pass through or whether the position can be used as a foothold of the biped robot, parameter optimization is performed on the foothold to obtain a target foothold sequence, such that the biped robot is enabled to complete a crossing task in view of terrain characteristics.

Description

Motion planning for bipedal robots Technical Field

[0001] The present invention relates to the technical field of legged robots, and in particular to motion planning of a bipedal robot. Background Art

[0002] Service robots have garnered considerable attention in recent years. These robots possess advanced mobility, perception, and manipulation capabilities, making them suitable for replacing traditional human labor in specific areas such as logistics, inspection, and household maintenance, effectively alleviating labor shortages in these sectors. Service robots can be broadly categorized into wheeled and legged types based on their configuration. Gait generation for bipedal robots has been a research hotspot in the field of legged robots.

[0003] At present, gait planning for bipedal robots is achieved by calculating the angles of each joint of the bipedal robot based on a model, that is, generating a time-based motion trajectory sequence of the joints so that the joints can accurately execute these trajectory sequences to complete the movement. For example, a human-like kinematic model and multiple center of mass trajectories are used to switch between gaits of different heights; or a penguin robot model is combined with the zero moment point (ZMP) trajectory method to generate a straight walking gait. Since there is an underactuated joint between the legs of the bipedal robot and the ground, the movement of this joint is uncontrollable, so the walking planning method based on the joint trajectory sequence is difficult to apply. In addition, in the gait planning method of the bipedal robot, it is assumed that the terrain on which the robot is located is flat. For terrain with fluctuating heights, the gait generated by the above method cannot be directly applied.

[0004] Summary of the Invention

[0005] In order to overcome the above-mentioned problems existing in the related art, this specification provides the following technical solutions.

[0006] According to a first aspect of the embodiments of this specification, a bipedal robot motion planning method is provided, comprising: obtaining a discrete map having terrain information corresponding to an area to be walked, determining terrain points in the discrete map that represent terrain points that allow the bipedal robot to pass through, and generating a preliminary landing point sequence based on the terrain points; generating a preliminary motion trajectory of the bipedal robot based on the discrete map; and optimizing the preliminary landing point sequence based on a preset objective function according to the preliminary motion trajectory and a kinematic model of the bipedal robot to generate a target landing point sequence, so that the bipedal robot can complete the motion by executing the target landing point sequence.

[0007] In one embodiment, the obtaining of a discrete map having terrain information corresponding to the area to be walked and generating a preliminary landing point sequence based on the terrain points may include: obtaining a discrete map having terrain information corresponding to the area to be walked and performing gradient calculation on the discrete map; taking terrain points in the discrete map whose gradient calculation results are less than or equal to the maximum landing height of the bipedal robot as passable points, and generating a first terrain map based on the passable points; and determining a preliminary landing point sequence for the bipedal robot to pass through the area to be walked based on the first terrain map.

[0008] In one embodiment, obtaining a discrete map having terrain information corresponding to the area to be walked and generating a preliminary foothold sequence based on the terrain points may further include: using terrain points in the discrete map for which the gradient calculation result is equal to zero as footholds of the bipedal robot, and updating the first terrain map based on the footholds.

[0009] In one embodiment, the method may further include: taking the terrain points in the discrete map whose gradient calculation results are greater than the maximum landing height as obstacle points that cannot be crossed by the bipedal robot; adding grid weights based on the terrain information of the obstacle points and the passable points, and constructing a two-dimensional second terrain map for subsequent motion trajectory planning.

[0010] In one embodiment, generating the preliminary motion trajectory of the bipedal robot based on the discrete map may include: obtaining a kinematic model of the wheeled robot; obtaining an initial position and a destination position of the bipedal robot; and performing trajectory planning processing based on the kinematic model of the wheeled robot in combination with the second topographic map, the initial position and the destination position to generate the preliminary motion trajectory of the bipedal robot.

[0011] In one embodiment, the method may further include: determining optimized size parameters and ranges of the bipedal robot, and establishing a kinematic model of the bipedal robot.

[0012] In one embodiment, determining the optimized size parameters and ranges of the bipedal robot and establishing the kinematic model of the bipedal robot may include: obtaining at least one of the motion step length, lateral movement distance, and trunk orientation angle of the bipedal robot; and optimizing the landing point positions of the two supporting feet of the bipedal robot based on a constraint relationship associated with at least one of the motion step length, the lateral movement distance, and the trunk orientation angle.

[0013] In one embodiment, the optimizing of the foothold positions of the two supporting feet of the bipedal robot according to the constraint relationship associated with at least one of the motion step length, the lateral movement distance, and the trunk orientation angle includes at least one of the following: optimizing the foothold positions of the two supporting feet of the bipedal robot so that the positional relationship of the two footholds conforms to the constraint relationship between the motion step length and the foothold range; optimizing the foothold positions of the two supporting feet of the bipedal robot so that the positional relationship of the two footholds conforms to the lateral movement distance; optimizing the sole orientation angles of the two supporting feet of the bipedal robot according to a preset magnification so that the ratio of the sum of the two sole orientation angles to the trunk orientation angle reaches the magnification.

[0014] The present invention also provides a bipedal robot motion planning device, which includes: a motion constraint module, which is used to obtain a discrete map with terrain information corresponding to the area to be walked, determine the terrain points in the discrete map that represent the bipedal robot allowed to pass, and generate a preliminary landing point sequence based on the terrain points; a trajectory generation module, which is used to generate a preliminary motion trajectory of the bipedal robot based on the discrete map; a sequence generation module, which is used to optimize the preliminary landing point sequence based on a preset objective function according to the preliminary motion trajectory and the kinematic model of the bipedal robot, and generate a target landing point sequence, so that the bipedal robot can complete the movement by executing the target landing point sequence.

[0015] The present invention also provides a bipedal robot motion planning device, which includes a memory, a processor, and a bipedal robot motion planning program stored in the memory and runnable on the processor. When the processor executes the bipedal robot motion planning program, the steps of the above-mentioned bipedal robot motion planning method are implemented.

[0016] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-described bipedal robot motion planning methods.

[0017] According to the bipedal robot motion planning method, apparatus, device, and storage medium of the embodiments of this specification, a preliminary foothold sequence is generated based on terrain points that allow the bipedal robot to pass through, represented in a discrete map containing terrain information corresponding to the area to be walked, and a preliminary motion trajectory of the bipedal robot is generated based on the discrete map. Based on a preset objective function, the preliminary foothold sequence is optimized according to the preliminary motion trajectory and the kinematic model of the bipedal robot to generate a target foothold sequence, so that the bipedal robot can more efficiently complete the task of traversing the area to be walked by executing the target foothold sequence in combination with the terrain characteristics. The target foothold sequence is obtained by utilizing the kinematic characteristics of the bipedal robot to discretize terrain with fluctuating heights, and optimizing the parameters of the bipedal robot's footholds based on the gradient information of the ground height in the discrete map (including determining whether the terrain points allow the bipedal robot to pass through or whether they can serve as footholds for the bipedal robot).

[0018] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] FIG1a is a flow chart of a bipedal robot motion planning method according to an embodiment of this specification.

[0020] FIG1 b is a flow chart of a bipedal robot motion planning method according to another embodiment of this specification.

[0021] FIG2 is a schematic diagram showing the movement of a biped robot according to an embodiment of the present specification.

[0022] 3a to 3c are schematic diagrams illustrating terrain discretization according to an embodiment of this specification.

[0023] FIG4 is a schematic top view of a foothold constraint of a bipedal robot according to an embodiment of this specification.

[0024] 5a to 5c are schematic diagrams showing planning of discrete landing points according to an embodiment of this specification.

[0025] FIG6 is a schematic diagram showing planning of discrete landing points according to another embodiment of this specification.

[0026] FIG7 is a schematic diagram of a motion planning device for a bipedal robot according to an embodiment of this specification.

[0027] FIG8 is a schematic diagram of a bipedal robot motion planning device according to an embodiment of this specification. DETAILED DESCRIPTION

[0028] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of the present invention.

[0029] Next, the embodiments of this specification are described in detail.

[0030] Fig. 1 is a flow chart of a method according to an exemplary embodiment of the present specification. As shown in Fig. 1 , the method includes the following steps.

[0031] In step 110, a discrete map with terrain information corresponding to the area to be walked is obtained, terrain points representing the passage of the bipedal robot are determined in the discrete map, and a preliminary foothold sequence is generated based on the terrain points.

[0032] In step 120 , a preliminary motion trajectory of the biped robot is generated based on the discrete map.

[0033] In step 130, based on a preset objective function, the preliminary foothold sequence is optimized according to the preliminary motion trajectory and the kinematic model of the bipedal robot to generate a target foothold sequence, so that the bipedal robot can complete the movement by executing the target foothold sequence.

[0034] This embodiment aims to: use known terrain information and the kinematic characteristics of the biped robot (the legged robot can traverse different terrain heights) to generate a series of footholds to achieve the movement task of the biped robot from a specified posture across the terrain to a target posture. Among them, pose is a general term for position and orientation, which can be divided into two-dimensional pose and three-dimensional pose. The two-dimensional pose includes horizontal position and vertical orientation, a total of three degrees of freedom information. The three-dimensional pose includes three-dimensional position and orientation, a total of six degrees of freedom information. In the present disclosure, the pose is a two-dimensional pose, specifically including the position information of the foothold of the biped robot and the orientation information of the torso.

[0035] As an example, the bipedal robot motion planning method can be applied to a bipedal robot motion planning system, which is applied to a bipedal robot motion planning device. The above steps are described in detail as follows.

[0036] In step 110, a discrete map with terrain information corresponding to the area to be walked is obtained, terrain points representing the passage of the bipedal robot are determined in the discrete map, and a preliminary foothold sequence is generated based on the terrain points.

[0037] Service robots can be roughly divided into wheeled and legged types based on their configuration. Among them, wheeled robots use the rotation of the chassis wheels as the main driving force; compared to wheeled robots, legged robots have higher mobility and can adapt to terrain with large height fluctuations (such as stairs, etc.). As shown in Figure 2, a bipedal robot can step up the stairs from position A and then go down the stairs to position B. Compared with wheeled robots, it can adapt to more walking scenarios. However, the structure of legged robots is more complex, and there are more unstable factors in the system itself, especially for bipedal robots. It is necessary to consider relevant constraints in the process of trajectory planning to ensure the safe execution of the overall mobile task.

[0038] Leveraging the kinematic characteristics of bipedal robots, the team discretized terrain with fluctuating heights, incorporating these characteristics to more efficiently complete traversal tasks. Specifically, a discrete map is generated based on the known terrain information of the area the bipedal robot will be traversing. The ground height gradient in the discrete map then determines whether the location is passable or suitable for the bipedal robot to rest on.

[0039] As an example, in a discrete map, if the height of the terrain change is greater than the maximum lift height of the bipedal robot, the robot cannot directly pass through that location. Conversely, if the height of the terrain change is less than the maximum lift height of the bipedal robot, the robot can pass through that location. To clearly illustrate the motion characteristics of a bipedal robot traversing different terrain heights, the following uses stairs as an example to illustrate the motion of a bipedal robot under different terrain changes.

[0040] For example, referring to Figure 3a, if the range of terrain heights is concentrated in (0, 0.10, 0.20), the terrain height of region C is 0 meters, the terrain height of region D is 0.10 meters, and the terrain height of region E is 0.20 meters, the maximum foot-lift height of the bipedal robot is 0.15 meters. Obviously, the bipedal robot cannot directly lift its foot from the ground with a terrain height of 0 meters in region C to the ground with a terrain height of 0.20 meters in region E (forming a step relative to region C). Therefore, the terrain discretization method will mark the boundary points of the above-mentioned two terrains that cannot be directly crossed in regions C and E as obstacle points. However, the bipedal robot can directly cross from the ground with a terrain height of 0 meters in region C to the ground with a terrain height of 0.10 meters in region D (also forming a step relative to region C). In this case, the terrain discretization method will mark the boundary points of the above-mentioned two terrains that can be directly crossed in regions C and D as traversable points. Both obstacle points and traversable points are used in the planning process of the preliminary motion trajectory. This means that a preliminary foothold range can be formed based on the traversable points. Subsequently, optimal footholds can be determined based on the bipedal robot's kinematic constraints and the foothold range, resulting in a preliminary foothold sequence. Specifically, the foothold orientations of two adjacent footholds can be constrained. This means that the orientation angles of the two adjacent footholds, as well as their average, must not deviate significantly from the body's orientation angle, thereby reducing the probability of joint motor failure in the mechanical structure.

[0041] As an example, step 110 may specifically include: in step 111, obtaining a discrete map with terrain information corresponding to the area to be walked, and performing gradient calculation on the discrete map; in step 112, taking the terrain points in the discrete map whose gradient calculation results are less than or equal to the maximum landing height of the bipedal robot as passable points, and generating a first terrain map based on the passable points; in step 113, determining a preliminary landing point sequence for the bipedal robot to pass through the area to be walked based on the first terrain map.

[0042] Get the maximum foot lift height h of the bipedal robot currently used to cross the area to be walked max , which represents the range of heights that a bipedal robot can cross. The maximum leg-lift height can be the actual leg-lift height of the bipedal robot, or a maximum leg-lift height given based on actual needs or safety considerations. It is understood that the actual leg-lift height should be between 0 meters and the maximum leg-lift height.

[0043] Obtain a terrain height map h(x,y) representing the area to be walked, where x and y represent horizontal position coordinates and h represents the terrain height at the corresponding horizontal position coordinates. Rasterize the actual map to obtain the corresponding discrete map h(i,j), and calculate the gradient of each pixel (i.e. terrain point) in the discrete map h(i,j).max The terrain points are marked as traversable points, which represent the range in which the bipedal robot can pass through these traversable points to form a preliminary foothold.

[0044] The traversable points determined in this manner form a first topographic map. The traversable points in the first topographic map serve as candidate landing points for the bipedal robot's movement. Combined with the kinematic constraints on the orientation of the soles of two adjacent landing points, a preliminary landing point sequence is generated. This preliminary landing point sequence also needs to be combined with the path the bipedal robot takes from its initial position to its target position to select the optimal target landing point sequence.

[0045] Step 110 may further include, at step 112a, using a terrain point in the discrete map where the gradient calculation result is zero as a landing point for the bipedal robot, and updating the first terrain map based on the landing point. Step 112a may be performed in parallel with step 112, or before or after step 112, as long as it is performed before step 113.

[0046] In a discrete map, the gradient is less than the maximum lift height h max The traversable points include both boundary points and non-boundary points on the two types of terrain. For a bipedal robot, landing a foot on the intersection of steps can easily lead to unbalanced and dangerous situations. Therefore, considering the stability of the bipedal robot, all boundary points are marked as non-landing points.

[0047] Exemplarily, terrain points with non-zero gradients are characterized as boundary points, i.e., marked as non-landing points, while terrain points with zero gradients are characterized as non-boundary points and marked as landing points. As described in Figures 3a to 3c, Figure 3a represents the original discrete map with terrain height information, where the terrain height of region C is 0 meters, the terrain height of region D is 0.10 meters, and the terrain height of region E is 0.20 meters. Through gradient calculation, it is determined that the portion surrounded by the dotted line in Figure 3b is the boundary area. When optimizing the landing point, the landing point cannot be located within the boundary area. Of course, the landing point can be located within the area other than the boundary area in Figure 3b.

[0048] Based on the first terrain map determined in the aforementioned manner, traversable points in the first terrain map are used as candidate landing points for the bipedal robot. Combined with the kinematic constraints on the orientation of the soles of two adjacent landing points, a preliminary landing point sequence is generated. This preliminary landing point sequence is then combined with the path the bipedal robot takes from its initial position to its target position to select the optimal target landing point sequence.

[0049] In step 120 , a preliminary motion trajectory of the biped robot is generated based on the discrete map.

[0050] Based on the relationship between the gradient calculated from the discrete map and the bipedal robot's maximum leg lift height, terrain points in the discrete map are marked as obstacles, unreachable areas, and passable areas, thereby generating a second terrain map and the first terrain map. The marked obstacle points in the second terrain map can be used as parameters for path planning.

[0051] As an example, the method may also include: taking the terrain points in the discrete map whose gradient calculation results are greater than the maximum landing height as obstacle points that cannot be crossed by the bipedal robot; adding grid weights based on the terrain information of the obstacle points and the passable points, and constructing a two-dimensional second terrain map for subsequent trajectory planning.

[0052] Calculate the gradient of each pixel point (i.e. terrain point) in the discrete map, and make the calculated gradient greater than the maximum foot lift height h of the bipedal robot. max Continuing with Figures 3a to 3c, by comparing the gradient of each terrain point in the discrete map, we can obtain the obstacle points in the area enclosed by the dotted curve in Figure 3c. These obstacle points represent locations that are difficult for the bipedal robot to cross and need to be avoided during subsequent trajectory planning.

[0053] Through the above embodiment, by combining the relevant parameters of the bipedal robot with the terrain information of the area to be walked, the position points or position areas that the bipedal robot can pass through are determined as a preliminary landing point sequence. Subsequently, the optimal landing point can be selected from the preliminary landing point sequence to generate a target landing point sequence.

[0054] As an example, step 120 may specifically include: in step 121, obtaining the kinematic model of the wheeled robot; in step 122, obtaining the initial posture and the target posture of the bipedal robot; in step 123, combining the second topographic map, the initial posture and the target posture, performing trajectory planning processing based on the kinematic model of the wheeled robot to generate a preliminary motion trajectory of the bipedal robot.

[0055] Given that the trajectory algorithms related to wheeled robots are relatively mature and widely used, the trajectory planning problem of legged robots can be borrowed from the trajectory planning methods of wheeled robots to a certain extent. For example, the trajectory planning method of wheeled robots can be used to generate a reference trajectory for legged robots. Specifically, the slope of each terrain grid is calculated using a discrete map, and excessive slopes (gradients greater than the maximum leg lift height h) are considered as the reference trajectory of legged robots. max ) is set as the obstacle point that the biped robot cannot cross, and the slope is moderate and can be crossed (the gradient is less than the maximum foot lift height h max, or with a gradient equal to zero) are set as traversable points for the bipedal robot, and corresponding network weights are set according to the slope to construct a two-dimensional second terrain map. It should be noted that the grid weight corresponding to the map grid acts as a penalty factor. For example, for terrain areas that are difficult to cross, a higher grid weight can be set in the corresponding map grid. Typically, the network weight is 0 or 1, with 1 indicating a non-traversable point and 0 indicating a traversable point.

[0056] The trunk motion model of the bipedal robot is abstracted and transformed into a virtual wheeled robot. The pose information is used to plan the overall motion trend of the bipedal robot, including the position information of the footholds in the second topographic map and the trunk orientation information. The kinematic model of the wheeled robot is shown below:

[0057] Among them, the linear velocity v is Within the interval, the unit is meter / second; d f is the maximum distance the biped robot moves forward; d b is the maximum distance the biped robot can move backward; T s The time it takes for the bipedal robot to move one step. The limit of the angular velocity ω is determined by the trunk rotation ability of the bipedal robot and is usually between 10 degrees / second and 20 degrees / second.

[0058] Then, the kinematic model of the wheeled robot is used in combination with Differential Dynamic Programming (DDP) to generate a sufficiently smooth posture curve connecting the initial posture and the target posture.

[0059] In the process of generating a preliminary motion trajectory for the bipedal robot by performing trajectory planning based on the wheeled robot's kinematic model, in conjunction with the second topographic map, the initial position, and the target position, the relevant parameters of the wheeled robot can be set to match the bipedal robot's motion parameters. For example, the maximum limit on the wheeled robot's linear velocity can be set based on the bipedal robot's maximum (forward or backward) speed, and the maximum limit on the wheeled robot's angular velocity can be set based on the bipedal robot's maximum turning speed. Thus, the trajectory constructed based on the virtual wheeled robot model after setting parameters can serve as a reference for the bipedal robot's center of mass movement and trunk orientation, that is, as the bipedal robot's preliminary motion trajectory for subsequent optimization of footholds.

[0060] In step 130, based on a preset objective function, the preliminary foothold sequence is optimized according to the preliminary motion trajectory and the kinematic model of the bipedal robot to generate a target foothold sequence, so that the bipedal robot can complete the movement by executing the target foothold sequence.

[0061] Using the initial motion trajectory as a reference trajectory for the bipedal robot to move from its initial position to its final position, and after defining the kinematic constraints of the bipedal robot, different target footfall sequences can be derived based on different optimization objectives. The bipedal robot can then walk from its starting point to its final position by executing this target footfall sequence.

[0062] Research has found that the motion characteristics of bipedal robots and wheeled robots have certain similarities. For example, without considering wheel slippage, the instantaneous linear velocity of a wheeled robot is consistent with the robot's orientation; similarly, the bipedal robot's motion ability is strongest in the direction of the robot's orientation. However, due to the discrete nature of its actual motion, the bipedal robot may adjust its lateral body movement based on the foothold at a certain moment; and in order to ensure the stability of the overall walking, the lateral movement distance of the bipedal robot must be guaranteed to be within a certain range. Therefore, as shown in Figure 1b, before optimizing the preliminary foothold sequence based on the preset objective function, the preliminary motion trajectory, and the kinematic model of the bipedal robot to generate a target foothold sequence, the method may also include: in step 130a, determining the optimized size parameters and range of the bipedal robot and establishing a kinematic model of the bipedal robot.

[0063] By establishing a kinematic model of the bipedal robot based on its structure, motion parameters, joint activity parameters, etc., the range of the bipedal robot during movement can be effectively limited, thereby ensuring that it can stably pass through areas with fluctuating terrain.

[0064] As an example, step 130a may specifically include: obtaining at least one of the bipedal robot's step length, lateral movement distance, and trunk orientation angle; and optimizing the foothold positions of the bipedal robot's two supporting feet based on a constraint relationship associated with at least one of the step length, lateral movement distance, and trunk orientation angle. Optimizing the foothold positions of the bipedal robot's two supporting feet based on a constraint relationship associated with at least one of the step length, lateral movement distance, and trunk orientation angle may include at least one of the following: optimizing the foothold positions of the bipedal robot's two supporting feet so that the positional relationship between the two footholds conforms to the constraint relationship between the step length and the foothold range; optimizing the foothold positions of the bipedal robot's two supporting feet so that the positional relationship between the two footholds conforms to the lateral movement distance; and optimizing the sole orientation angles of the bipedal robot's two supporting feet based on a preset magnification so that the ratio of the sum of the two sole orientation angles to the trunk orientation angle reaches the preset magnification.

[0065] Specifically, first of all, it can be understood that the distance that a biped robot steps forward or backward needs to be kept within a certain range. In view of this, the present invention proposes a kinematic constraint for the displacement between adjacent footholds of a biped robot. For example, referring to FIG4 , the positions of two adjacent footholds of a biped robot are (x p ,y p ,θ p ),(x n ,y n ,θ n ), the displacement between two adjacent landing points must satisfy the following relationship (1): b ≤(x n -x p )cosθ+(y n -y p )sinθ≤d f (1)

[0066] Among them, the displacement of the front and rear feet is regarded as a vector, d f represents the maximum forward step length allowed by the bipedal robot, d b represents the maximum backward step length allowed by the bipedal robot, and θ represents the trunk orientation angle of the bipedal robot. This expression (1) constrains the step length range of the bipedal robot along the body orientation, that is, the distance between the front and rear footholds of the bipedal robot along the trunk orientation is within a certain range.

[0067] Secondly, it is understood that the lateral step of the bipedal robot cannot deviate too much from the shoulder width, otherwise it will cause unnecessary lateral vibration of the trunk and reduce the stability of walking. In view of this, the present invention proposes a kinematic constraint for the lateral movement distance of the bipedal robot, that is, the distance between the front and rear footholds of the bipedal robot perpendicular to the direction of the trunk should be kept close to the shoulder width. For example, referring to Figure 4, the postures of the two adjacent footholds of the bipedal robot are (x p ,y p ,θ p ),(x n ,y n ,θ n ), the lateral movement range of the biped robot must satisfy the following relationship (2): hd s ≤(-1) 1-δ (x n -x p )sinθ+(-1) δ (y n -y p )cosθ≤h+d s (2)

[0068] Among them, the displacement of the front and rear feet is regarded as a vector, d s represents the maximum lateral deviation allowed by the bipedal robot, h represents the distance between the two supporting feet of the bipedal robot, and θ represents the trunk orientation angle of the bipedal robot. δ indicates whether the foothold corresponds to the left or right foot support, which is also referred to as the phase of the foothold. When the foothold is the left foot support, the corresponding δ value is 0, and when the foothold is the right foot support, the corresponding δ value is 1. This expression (2) constrains the lateral movement range of the bipedal robot and can effectively avoid unnecessary trunk swing caused by excessive lateral movement.

[0069] In addition, it is understood that the orientation angles of two adjacent footholds and their average value should not deviate too much from the orientation angle of the body. In view of this, the present invention proposes a kinematic constraint for the orientation angles between adjacent footholds of a biped robot. For example, referring to FIG4 , the postures of two adjacent footholds of a biped robot are (x p ,y p ,θ p ),(x n ,y n ,θ n ), the orientation of the soles of the two footholds must satisfy the following relationship (3): |θ n -θ|≤θ max |θ p -θ|≤θ max (3)

[0070] Among them, θ p Indicates the current landing point (x p ,y p )’s landing angle, θ n Indicates the next landing point (x n ,y n ) is the footfall angle, θ is the trunk angle of the biped robot, and θ max The maximum steering angle allowed by the biped robot is . This expression (3) constrains the orientation of the soles of two adjacent footholds of the biped robot, that is, it requires that the orientation angles of the footholds of two adjacent feet and their average value should not deviate too much from the orientation angle of the body.

[0071] The above constraints can effectively reduce the search space for the bipedal robot's foothold, thereby improving the efficiency of subsequent foothold optimization. For example, as shown in Figure 4, the dotted area F where the right supporting foot is located is the range where the right supporting foot is allowed to land.

[0072] Referring to Figures 5a to 5c, the bipedal robot needs to reach the destination posture B from the initial posture A in Figure 5a, passing through three areas of different terrain heights, CDE. Based on the above-mentioned kinematic model of the bipedal robot and the preliminary motion trajectory shown in Figure 5b, the foothold sequence is optimized and generated. The initial value of the foothold sequence is the aforementioned preliminary foothold sequence, including the number of footholds and the horizontal position coordinates, orientation, and phase of each foothold (corresponding to left foot support or right foot support). An example can be shown in Figure 5c. Therefore, the following optimization problem is constructed to generate the foothold sequence:

[0073] Where n represents the number of footholds in the foothold sequence; i is used to identify the i-th foothold in the foothold sequence; (x i ,y i ,θ i ) represents the i-th landing point in the landing point sequence, (x i ,y i ) represents the horizontal position coordinate of the foothold, θ i Indicates the landing angle of the landing point; represents the optimized i-th landing point, Indicates the horizontal position coordinates of the optimized landing point, represents the orientation angle of the optimized foothold, Indicates the phase of the optimization foothold; argmin means solving the function The set of optimized landing points that obtain the minimum value; λ0 represents the degree of execution time requirement of the task to be executed, hereinafter referred to as time weight; λ iIt represents the sensitivity of the task to the state of the i-th landing point, hereinafter referred to as the state weight.

[0074] Assuming the task execution time for each landing point remains constant, the overall task execution time for the complete landing point sequence will be proportional to the number of landing points. Therefore, when the time weight λ0 increases, it indicates that the task requires the shortest possible execution time; when the value of λ0 decreases, it indicates that the task is less sensitive to execution time. In the extreme case of λ0 being 0, it means that the execution time of the task is completely indifferent.

[0075] In addition, since the state of the foothold will also affect the success rate of task execution and the safety of the robot, some tasks are also sensitive to different states of different footholds. Generally, the first and last steps of a bipedal robot represent the starting and braking stages of its movement. These two stages involve the accumulation and dissipation of the overall trunk momentum, and the probability of instability is high. Therefore, when the task to be performed has high requirements for stability, the state weights λ1 and λ2 of the first and last footholds will be increased accordingly. n The state weights λi (i=2,3,....,n-1) of the remaining intermediate landing points can use roughly the same values.

[0076] Since the foothold of the biped robot involves the left and right supporting feet, the present invention proposes the concept of the phase of the foothold, which is exemplarily expressed as the following formula (4): δ i ∈{0,1} (4)

[0077] Here, formula (4) uses a binary variable to represent the phase of the foothold (i.e., whether it corresponds to left foot support or right foot support), where 0 represents left foot support and 1 represents right foot support.

[0078] The present invention also proposes a kinematic constraint for the phase of two adjacent landing points, which is expressed as the following formula (5): δ i+1 =1-δ i (5)

[0079] Among them, δ i represents the phase of the i-th landing point, δ i+1 The formula (5) constrains the phase of the two adjacent footholds to change, that is, the bipedal robot can only move by alternating between the left and right feet.

[0080] As mentioned above, in order to avoid the risk of imbalance caused by the bipedal robot stepping on a terrain point with a non-zero gradient, the landing point of the bipedal robot should be a terrain point with a zero gradient, which can be specifically expressed as the following formula (6): (x i ,y i )∈M feasible (6)

[0081] Among them, (x i ,y i ) represents the horizontal position coordinate of the i-th foothold, M feasible represents the set of footholds of the bipedal robot in the first terrain map. Formula (6) constrains the correlation between the footholds and the discretized terrain, that is, the foot of the bipedal robot cannot step on the obstacle point of the first terrain map to prevent the foot from being suspended in the air.

[0082] Based on the above formulas (1)-(3), (x p ,y p ) is the horizontal position coordinate of the current landing point, (x n ,y n ) is the horizontal position coordinate of the next landing point. i is used to identify the current landing point and i+1 is used to identify the next landing point. The kinematic constraints of the bipedal robot can be summarized as follows:

[0083] Among them, θ i represents the landing angle of the i-th landing point; θ i+1 represents the landing angle of the i+1th landing point; The bipedal robot's trunk heading angle, determined based on the reference trajectory, can be obtained by solving for the angle of the tangent line to the reference trajectory at the intersection of the reference trajectory and the line connecting two adjacent footholds. This formula constrains the bipedal robot's foothold range, ensuring that the next foothold is within a limited range relative to the current foothold. This effectively narrows the search range.

[0084] When the above optimization problem is used to generate a target landing point sequence, the initial and termination conditions must also be met. That is, the position and orientation of the first landing point must be consistent with the initial pose, and the position and orientation of the last landing point must be consistent with the target pose, as shown in Figures 2 and 5a.

[0085] Furthermore, the weighted parameters in the optimization objective can be selected based on the actual task scenario. For example, if a bipedal robot is required to reach the target pose in the shortest possible time, the number n of foothold sequences can be used as the objective parameter. Alternatively, if a bipedal robot is required to move to the target pose as smoothly as possible, the sum of the weights of the footholds on a discrete terrain map can be used as the objective parameter. By rationally adjusting the weights in the objective function, the legged robot's task performance can be effectively improved.

[0086] In the following, the sum of the weights of the landing points on the discrete map is taken as the target parameter as an example.

[0087] When λi=1,2,...,n is zero, the overall objective function is the number of footholds. This means the bipedal robot must be controlled to reach its final position from its initial position in the minimum number of steps. For example, using the example shown in Figures 5a to 5c, the bipedal robot's path should be to turn left, step onto the steps between areas C and D, walk straight, descend the steps between areas D and C, turn right, and reach its final position. This means the robot must fully utilize the terrain for movement.

[0088] When the overall objective function is the height of the foothold sequence, the optimal solution is to minimize the number of steps required. For example, using Figures 5a to 5c as an example, the planned footholds do not involve ascending or descending stairs. Instead, they maintain a constant movement from the initial position, with the left foot on the ground and the right foot on the stairs, to the desired position.

[0089] In some other embodiments, referring to FIG6 , the kinematic constraints of the biped robot are as follows: The maximum forward step length d is determined according to the structure and motion characteristics of the biped robot. f is 0.5 meters, and the maximum backward step length d b The maximum lateral deviation distance d is 0.3 meters, the distance h between the two support legs is 0.5 meters, and the maximum lateral deviation distance d is 0.3 meters. s is 0.05 meters, the maximum rotation angle of the support leg is θ max It is 10 degrees.

[0090] Set the initial pose (0, 0, 0) and the target pose The preliminary motion trajectory of the biped robot obtained based on the wheeled robot model is: as a reference trajectory.

[0091] Continuing with FIG6 , it is assumed that the restricted area of ​​the foothold obtained by terrain discretization can be represented as the gray area G[0.6,0.7]×[0,0.6] in FIG6 .

[0092] According to plane geometry, given the positions of the two front and rear footholds (x1, y1) and (x2, y2), the angle of the intersection of their connecting line and the reference trajectory can be calculated using the following analytical expression: a=x1-x2; b=y1-y2; c=x1y2-x2y1.

[0093] The reference trajectory length is set to L, and the fixed step length of each step is The average of the foot sole orientations of two adjacent footholds of the biped robot is equal to the trunk orientation angle of the biped robot. The preliminary foothold sequence of the above embodiment can be obtained by the following formula: δ1∈{0,1}, y1=0.25·(-1) δ1 , δ2=1-δ1.

[0094] The initial landing point sequence calculated by the above formula does not necessarily meet the boundary conditions for termination. It is necessary to combine the kinematic constraints of the bipedal robot, the initial motion trajectory, and the preset objective function, and gradually determine the target landing point sequence that meets the boundary conditions for termination through the IQP boundary value algorithm.

[0095] Specifically, assuming that the number of footholds n is 2, the parameters that need to be optimized are the posture and support phase of the first foothold (x1, y1, θ1, δ1), and the posture and support phase of the second foothold (x2, y2, θ2, δ2). Substituting the above parameters into the above constraint formulas (1)-(6), the corresponding relationship is expressed as follows: δ1∈{0, 1}, δ2=1-δ1,

[0096] x1<0.6 or x1>0.7 or y1>0.6,

[0097] x2<0.6 or x2>0.7 or y2>0.6.

[0098] Based on the above constraints, it can be determined that the quantities that need to be optimized are the support phase of the first foothold and the positions of the two footholds. Other relevant quantities can be calculated based on the above constraints. In some embodiments, the objective function expressed as follows can be used: M = (θ1-θ2) 2 .

[0099] The physical meaning of this objective function is to minimize the difference in the plantar orientation between the two center footholds. This is because, in practical applications, if the angle difference between the two supporting legs of a bipedal robot is too large, it can easily cause some joint motors to approach mechanical limits and also cause self-collision during leg swinging. This objective function can significantly reduce the probability of these problems, thus protecting the robot.

[0100] In some embodiments, the following objective function may also be used:

[0101] The physical meaning of this objective function is to minimize the lateral deviation between adjacent footholds. This is because, during bipedal robot motion, lateral deviation often leads to significant trunk vibration. If the trunk's center of gravity is too high, this can cause instability and lead to the robot tipping over. This objective function can effectively reduce the magnitude of lateral deviation and improve motion stability.

[0102] In some embodiments, an objective function targeting the number of footholds n can also be used. Specifically, the movement of a bipedal robot can often be assumed to take the same amount of time for each step. Minimizing the number of footholds n means the fewest steps, and therefore the shortest movement time. In other words, the physical meaning of the objective function targeting the minimum number of footholds n is to reduce the time required for the bipedal robot to move from its initial position to its destination position. This means that the robot must fully utilize its movement capabilities to reach its destination position in the shortest possible time.

[0103] Based on the objective function, the foothold sequence optimization problem is solved to generate the target foothold sequence. The specific process can be briefly described as follows: When the number of footholds is determined (such as the case of n=2 mentioned above), the value of the support leg phase corresponding to the first foothold can be 0 or 1, and these two cases can be calculated separately. In particular, when the value of the support leg phase is determined, the kinematic constraint of the above-mentioned bipedal robot will become a nonlinear constraint vector, and the corresponding objective function is usually also nonlinear, so the iterative quadratic programming (IQP) method can be used to solve it to obtain the local optimal value. Then, the optimal target value is calculated for the two cases where the support leg phase corresponding to the first foothold is 0 and 1, and the support angle phase with the lower optimal target value is taken as the final solution.

[0104] However, in reality, the number of footholds is often uncertain, so it's necessary to find a number of footholds that satisfies the overall constraints, also known as finding a solution. Furthermore, for search efficiency reasons, a binary search can be performed on the number of footholds to find a solution in a relatively small number of attempts, assuming the optimal solution exists. For example, since the foothold sequence must include at least the first foothold, i.e., the initial pose, and the last foothold, i.e., the destination pose, the lower limit of the number of footholds is clearly 2. We can first check whether a solution exists with two footholds. If a solution exists with two footholds, this means the bipedal robot can reach the destination pose directly from the initial pose (which is essentially unlikely). If a solution does not exist with two footholds, the number of footholds can be multiplied by 2 to obtain a new number of footholds, 4, and the search can be continued to see whether a solution exists with this new number of footholds. If a solution still does not exist with this new number of footholds, a new number of footholds is generated and the search continues. Repeat this process until the number of footholds corresponding to the situation where a solution exists is found, and this number of footholds is determined as the minimum number of footholds that can meet the overall constraints; or, until the number of new footholds generated is too large (for example, exceeds a predetermined threshold) and the search process for the number of footholds is exited.

[0105] Alternatively, if it is predetermined that there is a certain number of footholds n that can satisfy the overall constraints, then a new number of footholds can be selected And check if there is a solution in this case. If the number of new landing points If there is a solution, a new number of footholds will be generated again. And check if there is a solution in this case. If the number of new landing points If a solution still exists, a new number of footholds is generated and the search continues. This process is repeated until the number of footholds corresponding to the case where no solution exists is found. The previous number of footholds corresponding to this number of footholds is determined as the minimum number of footholds that can satisfy the existence of a solution, that is, the minimum number of footholds that satisfies the overall constraint.

[0106] It should be noted that, the above describes how to find the number of landing points using the binary method as an example, but those skilled in the art should understand that the binary method is adopted only for the sake of search efficiency, and the present disclosure should not be limited to this. In fact, any other search method known to those skilled in the art (for example, gradually increasing from the lower limit of the number of landing points with a specific step size, or gradually decreasing from the number of landing points with a specific step size for which a solution exists to generate a new number of landing points) is also applicable to the embodiments of the present disclosure. In addition, if the objective function includes the number of landing points and other related parameters, the minimum value of the objective function can be recorded during the entire search for the number of landing points, and the minimum value of the objective function and the corresponding solution can be reported when the minimum number of landing points is found.

[0107] The bipedal robot motion planning method provided by the present invention generates a preliminary foothold sequence based on terrain points that allow the bipedal robot to pass through, represented in a discrete map containing terrain information corresponding to the area to be walked, and generates a preliminary motion trajectory of the bipedal robot based on the discrete map. This allows the preliminary foothold sequence to be optimized based on a preset objective function, the preliminary motion trajectory, and the kinematic model of the bipedal robot to generate a target foothold sequence, so that the bipedal robot can more efficiently complete the task of traversing the area to be walked by executing the target foothold sequence in combination with the terrain characteristics. The kinematic characteristics of the bipedal robot are utilized to discretize terrain with fluctuating heights, and the foothold parameters of the bipedal robot are optimized based on the gradient information of the ground height in the discrete map (including determining whether a terrain point allows the bipedal robot to pass through or whether it can serve as a foothold for the bipedal robot), thereby obtaining a target foothold sequence.

[0108] Based on the same application concept as the above-mentioned method, an embodiment of the present invention further provides a bipedal robot motion planning device, as shown in FIG7 . The device includes: a motion constraint module 702 for obtaining a discrete map containing terrain information corresponding to the area to be traveled, determining terrain points in the discrete map that represent terrain points that the bipedal robot is allowed to pass through, and generating a preliminary landing point sequence based on the terrain points; a trajectory generation module 704 for generating a preliminary motion trajectory of the bipedal robot based on the discrete map; and a sequence generation module 706 for optimizing the preliminary landing point sequence based on a preset objective function, the preliminary motion trajectory, and the kinematic model of the bipedal robot to generate a target landing point sequence, so that the bipedal robot can complete the motion by executing the target landing point sequence.

[0109] Optionally, the motion constraint module 702 is further used to: obtain a discrete map with terrain information corresponding to the area to be walked, and perform gradient calculation on the discrete map; use terrain points in the discrete map whose gradient calculation results are less than or equal to the maximum landing height of the bipedal robot as passable points, and generate a first terrain map based on the passable points; and determine a preliminary landing point sequence for the bipedal robot to pass through the area to be walked based on the first terrain map.

[0110] Optionally, the motion constraint module 702 is further configured to: use a terrain point in the discrete map where a gradient calculation result is equal to zero as a landing point, and update the first terrain map based on the landing point.

[0111] Optionally, the motion constraint module 702 is also used to: use the terrain points in the discrete map whose gradient calculation results are greater than the maximum landing height of the bipedal robot as obstacle points that cannot be crossed by the bipedal robot; add grid weights according to the terrain information of the obstacle points and the passable points, and construct a two-dimensional second terrain map for subsequent motion trajectory planning.

[0112] Optionally, the trajectory generation module 704 is also used to: obtain a kinematic model of a wheeled robot; obtain an initial position and a destination position of the bipedal robot; and perform trajectory planning based on the kinematic model of the wheeled robot in combination with the second topographic map, the initial position and the destination position to generate a preliminary motion trajectory of the bipedal robot.

[0113] Optionally, the sequence generation module 706 is further configured to determine optimized size parameters and ranges of the bipedal robot and establish a kinematic model of the bipedal robot.

[0114] Optionally, the sequence generation module 706 is also used to: obtain at least one of the motion step length, lateral movement distance, and trunk orientation angle of the bipedal robot; and optimize the landing point positions of the two supporting feet of the bipedal robot based on a constraint relationship associated with at least one of the motion step length, the lateral movement distance, and the trunk orientation angle.

[0115] The implementation process of the functions and effects of each module / submodule / unit in the above-mentioned device is specifically described in the implementation process of the corresponding steps in the above-mentioned method, and the same technical effects can be achieved, so it will not be repeated here.

[0116] Corresponding to the aforementioned method embodiments, this specification also provides embodiments of an apparatus and a terminal to which it is applied.

[0117] The embodiments of the bipedal robot motion planning device of this specification can be applied to computer devices, such as servers or terminal devices. The device embodiments can be implemented by software, or by hardware or a combination of software and hardware. Taking software implementation as an example, as a device in a logical sense, it is formed by the processor of the bipedal robot motion planning in which it is located reads the corresponding computer program instructions in the non-volatile memory into the memory and runs them. From the hardware level, Figure 8 shows a hardware structure diagram of the computer device where the bipedal robot motion planning device 831 according to the embodiment of this specification is located. According to the actual function of the computer device, in addition to the processor 810, memory 830, network interface 820, and non-volatile memory 840 shown in Figure 8, other hardware may also be included according to the actual function of the computer device, which will not be described in detail.

[0118] The foregoing description of this specification describes specific embodiments. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in an order different from that described in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or the sequential order to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0119] Other embodiments of the present invention will readily occur to those skilled in the art upon consideration of the present invention and practice of the invention claimed herein. This specification is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of this specification and include common knowledge or customary techniques in the art not claimed herein. The description and examples are to be considered as exemplary only, with the true scope and spirit of the present invention being indicated by the following claims.

[0120] It should be understood that the present description is not limited to the exact construction that has been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present description is limited only by the appended claims.

[0121] The above description is only a preferred embodiment of this specification and is not intended to limit this specification. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of this specification should be included in the scope of protection of this specification.

Claims

1. A motion planning method for a biped robot, comprising: Obtaining a discrete map with terrain information corresponding to the area to be walked, determining the terrain points in the discrete map that represent the terrain through which the biped robot is allowed to pass, and generating a preliminary sequence of foothold points based on the terrain points; Generating a preliminary motion trajectory of the biped robot based on the discrete map; Based on a preset objective function, optimizing the preliminary sequence of foothold points according to the preliminary motion trajectory and the kinematic model of the biped robot to generate a target sequence of foothold points for the biped robot to complete the motion by executing the target sequence of foothold points.

2. The motion planning method for a biped robot according to claim 1, wherein, The obtaining of the discrete map with terrain information corresponding to the area to be walked and generating a preliminary sequence of foothold points based on the terrain points includes: Obtaining a discrete map with terrain information corresponding to the area to be walked and performing gradient calculation on the discrete map; Taking the terrain points in the discrete map with gradient calculation results less than or equal to the maximum foothold height of the biped robot as passable points and generating a first topographic map based on the passable points; Determining a preliminary sequence of foothold points for the biped robot to pass through the area to be walked according to the first topographic map.

3. The motion planning method for a biped robot according to claim 2, wherein, The obtaining of the discrete map with terrain information corresponding to the area to be walked and generating a preliminary sequence of foothold points further includes: Taking the terrain points in the discrete map with gradient calculation results equal to zero as the foothold points of the biped robot and updating the first topographic map based on the foothold points.

4. The motion planning method for a biped robot according to claim 2 or 3, wherein, The method further includes: Taking the terrain points in the discrete map with gradient calculation results greater than the maximum foothold height as the obstacle points that the biped robot cannot cross; Adding grid weights according to the terrain information of the obstacle points and the passable points to construct a two-dimensional second topographic map.

5. The motion planning method for a biped robot according to claim 4, wherein, The generating of the preliminary motion trajectory of the biped robot based on the discrete map includes: Obtaining the kinematic model of a wheeled robot; Obtaining the initial pose and the target pose of the biped robot; Combining the second topographic map, the initial pose and the target pose, and performing trajectory planning processing based on the kinematic model of the wheeled robot to generate the preliminary motion trajectory of the biped robot.

6. The motion planning method for a biped robot according to any one of claims 1 to 5, wherein, The method further includes: Determining the optimized dimension parameters and ranges of the biped robot and establishing the kinematic model of the biped robot.

7. The motion planning method for a biped robot according to claim 6, wherein, The determining of the optimized dimension parameters and ranges of the biped robot and establishing the kinematic model of the biped robot includes: Obtain at least one of the movement step length, lateral movement distance, and torso orientation angle of the biped robot; Optimize the landing positions of the two support feet of the biped robot according to the constraint relationship associated with at least one of the movement step length, the lateral movement distance, and the torso orientation angle.

8. The biped robot motion planning method according to claim 7, characterized in that the optimizing the landing positions of the two support feet of the biped robot according to the constraint relationship associated with at least one of the movement step length, the lateral movement distance, and the torso orientation angle includes at least one of the following: Optimize the landing positions of the two support feet of the biped robot so that the positional relationship between the two landing positions conforms to the constraint relationship between the movement step length and the landing range; Optimize the landing positions of the two support feet of the biped robot so that the positional relationship between the two landing positions conforms to the lateral movement distance; According to a preset magnification, optimize the sole orientation angles of the two support feet of the biped robot so that the ratio of the sum of the two sole orientation angles to the torso orientation angle reaches the magnification.

9. A biped robot motion planning device, comprising: A motion constraint module, configured to obtain a discrete map with terrain information corresponding to the area to be walked, determine the terrain points in the discrete map that represent the terrain points where the biped robot is allowed to pass, and generate a preliminary landing point sequence based on the terrain points; A trajectory generation module, configured to generate a preliminary motion trajectory of the biped robot based on the discrete map; A sequence generation module, configured to optimize the preliminary landing point sequence based on a preset objective function according to the preliminary motion trajectory and the kinematic model of the biped robot, and generate a target landing point sequence for the biped robot to complete the motion by executing the target landing point sequence.

10. A biped robot motion planning device, comprising: A memory, A processor, and A biped robot motion planning program stored on the memory and executable on the processor, wherein when the processor executes the biped robot motion planning program, the steps of the biped robot motion planning method according to any one of claims 1 to 8 are implemented.

11. A computer-readable storage medium, on which a biped robot motion planning program is stored, and when the biped robot motion planning program is executed, the steps of the biped robot motion planning method according to any one of claims 1 to 8 are implemented.

Images