Dynamic response path planning method and system for legged robot

Abstract

The invention provides a dynamic response path planning method and system for a legged robot. The method comprises the following steps: acquiring position information of the legged robot and sensing information of a surrounding environment; according to the current position and the sensing information, planning a reference path for reaching the target point, and according to the reference path, constructing a dynamic obstacle avoidance convex area; and predicting the future state of the legged robot according to a plurality of reference points selected on the reference path, constructing and solving an optimization problem containing dynamic obstacle avoidance convex region constraints by using a model prediction control algorithm according to a prediction result, generating an optimal control action, and converting the optimal control action into a control instruction. According to the method, the problem of dynamic response path planning of the legged robot under the condition of dynamic change of environments such as polling or dense obstacles can be solved, and efficient path planning and rapid dynamic response in complex and dynamic environments are realized.

Description

Technical Field

[0001] The present invention belongs to the field of robot path planning, and particularly relates to a dynamic response path planning method and system for a legged-footed robot. Background Art

[0002] The statements in this part only provide background technical information related to the present invention and do not necessarily constitute prior art.

[0003] In some scenarios, such as in autonomous inspection operations, factors such as obstacles, terrain changes, and dynamic targets in a complex dynamic environment pose severe challenges to the path planning and control of legged-footed robots. Currently, algorithms usually divide path planning and control into two independent stages: path planning is used to generate a global path, and the control module is responsible for following along the planned path. However, path planning usually needs to search the entire environment to calculate the optimal path, and the whole process is computationally complex and time-consuming. Especially when the environment changes dynamically or there are dense obstacles, it will cause system delays and affect the response ability of legged-footed robots in a dynamic environment.

[0004] In addition, there is a lack of real-time linkage between the path planning and control modules. Before the planned path is transmitted to the control module for execution, when the obstacles in the environment suddenly change, some methods of independent path planning and control cannot quickly re-plan the path. As a result, the legged-footed robot may fall into a local optimal solution or is prone to collisions in narrow and complex areas, reducing the efficiency and reliability of the inspection operation. Summary of the Invention

[0005] In order to solve the above problems, the present invention proposes a dynamic response path planning method and system for a legged-footed robot. The present invention can solve the problem of dynamic response path planning of legged-footed robots in cases of environmental dynamic changes or dense obstacles such as inspections, and realizes efficient path planning and fast dynamic response in complex and dynamic environments.

[0006] According to some embodiments, the present invention adopts the following technical solutions:

[0007] A dynamic response path planning method for a legged-footed robot, comprising the following steps:

[0008] Obtain the position information of the legged-footed robot and the perception information of the surrounding environment;

[0009] According to the current position and perception information, plan a reference path to the target point, and construct a dynamic obstacle avoidance convex region according to the reference path;

[0010] Predict the future state of the legged robot based on multiple reference points selected on the reference path. According to the prediction results, use the model predictive control algorithm to construct and solve an optimization problem with dynamic obstacle avoidance convex region constraints, generate the optimal control action, and convert the optimal control action into a control instruction.

[0011] As an alternative implementation, the process of obtaining the perception information of the surrounding environment includes obtaining the point cloud information and spatial attitude information of the surrounding environment.

[0012] As an alternative implementation, the process of planning the reference path to the target point based on the current position and perception information includes using a heuristic fast path search algorithm to determine the search path according to the map topology of the surrounding environment, and omitting unnecessary nodes during the search process.

[0013] As a further defined implementation, create a first list and a second list. The first list is used to store the nodes found and to be processed during the search process, and the second list is used to store the processed nodes;

[0014] Conduct the search, add the current position or the starting point of the legged robot to the first list, select the node with the smallest heuristic function from the first list as the current node and move it to the second list, and during the search, perform a straight-line search first according to the directions where the starting point and the target point are located, then perform an oblique search, and when encountering a jump point or an obstacle, complete the search in the current direction and add the jump point to the first list;

[0015] Repeat the above search process until the target point is added to the first list or the first list is empty, and find the reference path with the smallest cost function value.

[0016] As an alternative implementation, the process of constructing the dynamic obstacle avoidance convex region according to the reference path includes regarding the reference path as a set of a series of discrete points, each discrete point being a path segment. For each path segment, construct a rectangle with the midpoint of the line segment as its center, the length of the rectangle being in the same direction as the line segment, and the width being perpendicular to the direction of the line segment. Detect whether the rectangle contains an obstacle. If it contains an obstacle, gradually reduce the width of the rectangle until the rectangle boundary just touches the obstacle;

[0017] Generate a more accurate convex polygon according to the adjusted rectangle.

[0018] For each side of the rectangle, detect whether it contacts the obstacle. If it contacts, generate a tangent plane in the normal direction;

[0019] Gradually expand the rectangle along the direction of the path segment, and at the same time add new obstacle points to generate tangent planes, and finally form a convex polygon.

[0020] As a further step, when passing through an area with a width less than a set value or an obstacle distribution density greater than a set value, the generated convex polygon is shrunk, and each edge of the convex polygon is shifted inward by a certain length in the normal direction to ensure that the legged robot is always within a safe range. The length of the inward shift is the semi-diagonal length of the legged robot.

[0021] As an alternative implementation, the process of predicting the future state of the legged robot based on multiple reference points selected on the reference path includes: sampling a series of reference points on the reference path, and using the positions of each reference point as the input of the model predictive control problem. The first reference position is the point closest to the current position of the legged robot.

[0022] Based on the positions of each reference point, the state of the legged robot in the future for a period of time is obtained based on model predictive control, including position, speed, and acceleration.

[0023] As an alternative implementation, using the model predictive control algorithm, the process of constructing and solving an optimization problem containing dynamic obstacle avoidance convex region constraints to generate the optimal control action includes:

[0024] The model constraints are described as:

[0025] X n = f(X n-1 , U n-1 )

[0026] where X follows the initial state of the legged robot, X0 = (p0, v0, a0) T , and p, v, a are the current position, speed, and acceleration of the legged robot respectively;

[0027] The dynamic constraints include the upper and lower limits of the speed and acceleration of the legged robot;

[0028] The dynamic obstacle avoidance convex region constraint is: S n p n - c ≤ 0;

[0029] where any line segment of the dynamic obstacle avoidance convex region is described as a i x + b i y + c i ≤ 0, and S n is a matrix, each row of which corresponds to the coefficients [a i , b i of one side of the polygon, and c is a vector, each element of which corresponds to c i ;

[0030] The model predictive control problem is formulated as a quadratic programming problem to solve for the optimal control action.

[0031] A dynamic response path planning system for a legged robot, comprising:

[0032] A data acquisition module configured to acquire the position information of the legged robot and the perception information of the surrounding environment;

[0033] An obstacle avoidance area construction module configured to plan a reference path to a target point according to the current position and perception information, and construct a dynamic obstacle avoidance convex area according to the reference path;

[0034] A path planning and control module configured to predict the future state of the legged robot based on multiple reference points selected on the reference path, and according to the prediction result, use a model predictive control algorithm to construct and solve an optimization problem including constraints of the dynamic obstacle avoidance convex area, generate an optimal control action, and convert the optimal control action into a control instruction.

[0035] A computer-readable storage medium for storing computer instructions, which, when executed by a processor, complete the steps in the above method.

[0036] An electronic device comprising a memory, a processor, and computer instructions stored on the memory and running on the processor, which, when run by the processor, complete the steps in the above method.

[0037] A legged robot comprising a robot body, a controller is provided on the robot body, computer instructions are stored on the controller to execute the steps in the above method, or the above system is provided on the robot body, or the above electronic device.

[0038] Compared with the prior art, the beneficial effects of the present invention are:

[0039] The present invention integrates the path planning and control problems in one framework by using model predictive control, avoiding the time delay between the legged robot first perceiving the environment for path planning and then following according to the path control.

[0040] The present invention adopts a heuristic fast path planning method to reduce the search time and space complexity, combined with model predictive control, which can achieve efficient path planning and fast dynamic response in a complex dynamic environment. At the same time, the proposed method for constructing a dynamic obstacle avoidance convex area can enable the robot to pass through narrow areas smoothly, effectively avoiding the situation where the robot passes close to obstacles, and ensuring that the robot is always in a safe state.

[0041] To make the above objects, features and advantages of the present invention more obvious and understandable, the following specific preferred embodiments are given, and in conjunction with the accompanying drawings, the detailed description is as follows. Description of the Drawings

[0042] The accompanying drawings forming a part of this invention are used to provide a further understanding of the invention. The schematic embodiments and descriptions thereof of the invention are used to explain the invention and do not unduly limit the invention.

[0043] Figure 1 It is a schematic flowchart of a dynamic response path planning method for a legged robot of an embodiment;

[0044] Figure 2 A schematic diagram of a reference path of an embodiment. Detailed implementation manners

[0045] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0046] It should be noted that the following detailed descriptions are all illustrative and are intended to provide further descriptions of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0047] It should be noted that the terms used herein are only for describing specific implementation manners and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular forms are also intended to include the plural forms. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0048] In the case of no conflict, the embodiments in the present application and the features in the embodiments can be combined with each other.

[0049] Embodiment 1

[0050] A dynamic response path planning method for a legged robot, as Figure 1 shown, includes the following steps:

[0051] S1: Use a laser sensor module to obtain real-time perception information of the surrounding environment.

[0052] In this embodiment, the solid-state lidar mid360 is used to obtain real-time perception information of the surrounding environment.

[0053] Of course, in other embodiments, other devices can also be selected to obtain the perception information of the surrounding environment, such as a 3D laser scanner, etc., which will not be elaborated here.

[0054] An inertial measurement unit (imu) can also be used to obtain imu data to determine the three-axis attitude angle (or angular rate) and acceleration of the robot.

[0055] S2: The positioning module processes the positioning using the laser inertial odometry method based on the point cloud data and imu data collected in real time, obtains the real-time position of the legged robot, and performs path planning and control based on the obtained position data and point cloud data.

[0056] Of course, in other embodiments, other algorithms can also be selected for positioning.

[0057] Similarly, in some embodiments, other devices can also be selected for robot positioning, such as ultrasonic navigation positioning, binocular positioning, or other laser positioning, etc., which will not be enumerated here.

[0058] When performing path planning and control, first plan a reference path to the target point according to the position of the legged robot and the surrounding environment; then construct a dynamic obstacle avoidance convex region in real time according to the reference path, and this region can be described by a series of linear inequalities; by sampling a series of positions on the reference path as inputs, predict the future state, construct and solve an optimization problem containing linear inequality constraints, so as to generate the optimal control action and convert it into an actual control command to precisely control the legged robot.

[0059] Specifically, it includes:

[0060] S31: The path planning method used here is the heuristic fast path search algorithm, which omits some unnecessary nodes in the search by using the topological structure of the map, thereby reducing the time and space complexity of the search.

[0061] The heuristic function is:

[0062] f(n) = g(n) + h(n) (1)

[0063] In the formula, f(n) is the heuristic function value, which is the minimum cost estimate from the initial state through state n to the target state, g(n) is the cost function, which is the minimum cost from the initial state to state n in the state space, and h(n) is the prediction function, which is the minimum estimated cost of the path from state n to the target state.

[0064] S311: Define the dynamic data structures of Open List and Close List. Open List is used to store the nodes found and to be processed during the search, and Close List is the processed nodes; e List is the processed nodes;

[0065] S312: Add the current position (starting point) of the legged robot to the Open List, g(n) = 0, and h(n) is the predicted value;

[0066] S313: Select the node with the minimum f(n) from the Open List as the current node and move it to the Close List;

[0067] S314: When searching, first perform a straight-line search in the up, down, and right directions according to the directions of the starting point and the target point (to reduce unnecessary searches), and then perform a diagonal search. When encountering a jump point or an obstacle, complete the search in the current direction and add the jump point to the Open List.

[0068] S315: Repeat S313 and S314 until the target point is added to the Open List or the Open List is empty, complete the iterative search, and find the reference path with the minimum cost value according to the cost function, as Figure 2 shown, the reference path is S->G1->G2->G3->G.

[0069] S32: Construction of a dynamic obstacle avoidance convex region, and perform real-time construction of a dynamic convex region according to the reference path to provide linear constraints for model prediction.

[0070] S321: The reference path can be represented as a set of a series of discrete points:

[0071]

[0072] where p i is a path point in path planning. The path can be divided into multiple line segments, and each line segment is composed of p i and p i+1 and is expressed as:

[0073]

[0074] S322: Model the legged robot as a rectangle. For the path segment L i , construct a rectangle with the midpoint of the line segment as the center, the length of the rectangle consistent with the direction of the line segment, and the width perpendicular to the direction of the line segment.

[0075] The midpoint of the line segment is:

[0076]

[0077] The direction vector and unit vector of the line segment:

[0078] d i =(x i+1 -x i , y i+1 -y i )

[0079]

[0080] The unit vector in the vertical direction:

[0081]

[0082] The length of the rectangle is:

[0083] l i = ||d i || + 2r s (7)

[0084] The width of the rectangle is:

[0085] w i = w r + 2r s (8)

[0086] Wherein, w r is the width of the legged robot, and r s is the safety threshold, which is equal to the width of the robot here, to ensure that the path segment does not touch the obstacle.

[0087] S323: Adjust the rectangle size. Detect whether the rectangle contains an obstacle. If it contains an obstacle, gradually reduce the width w of the rectangle i until the rectangle boundary is just tangent to the obstacle, ensuring that w i is greater than 1.5w r .

[0088] S324: Polygon generation. Generate a more accurate convex polygon according to the adjusted rectangle, mainly by adding tangent planes to ensure the movement safety of the legged robot.

[0089] S325: Tangent plane generation. For each side of the rectangle, detect whether it touches the obstacle. If it touches, generate a tangent plane in the normal direction.

[0090] Let a point p on the side b touch the obstacle O j , then the tangent plane equation is:

[0091]

[0092] Wherein, is the normal direction of the rectangle side.

[0093] S326: Inflate the polygon. Gradually inflate the rectangle along the path segment direction, and at the same time add obstacle points to generate tangent planes, and finally form a convex polygon.

[0094] S327: Shrinkage processing. When passing through a narrow area or in a dense obstacle situation, it is necessary to shrink the generated polygon. Each edge of the polygon is shifted inward by r c in the normal direction to ensure that the legged robot is always within the safe range. The shrinkage distance rc is the semi-diagonal length of the legged robot:

[0095]

[0096] where l r is the length of the legged robot.

[0097] S33: Model Predictive Control. By sampling a series of positions on the reference path as inputs, predicting future states, constructing and solving an optimization problem with linear inequality constraints, so as to generate optimal control actions.

[0098] S331: Sample a series of reference positions p refn on the reference path, and these positions serve as the inputs to the model predictive control problem. The first reference position p ref1 is the point closest to the current position of the legged robot, and the last position p refn , p refn = v r ·N·Δt, where N is the prediction step number, Δt is the time step, and v r is the reference speed. Vr is the reference speed.

[0099] S332: Using these reference positions, use Model Predictive Control (MPC) to predict the states of the legged robot in the future for a period of time, including position, speed, and acceleration.

[0100] S333: Construct an optimization problem. Model Predictive Control mainly includes model constraints, dynamic constraints, and dynamic obstacle convex region constraints.

[0101] The model constraints are described as:

[0102] X n = f(X n-1 , U n-1 ) (11)

[0103] where X follows the initial state of the legged robot, X0 = (p0, v0, a0) T , specifically as follows:

[0104]

[0105] where p, v, a, j are the current position, speed, acceleration, and jerk of the legged robot respectively.

[0106] The dynamic constraints are described as:

[0107]

[0108] where i = j = x, y. The current speed v iLess than or equal to the maximum speed v max ;

[0109] The current acceleration a i Less than or equal to the maximum acceleration a max .

[0110] The dynamic convex region constraint is described as:

[0111] S n p n -c ≤ 0 (14)

[0112] Where any line segment of the dynamic obstacle avoidance convex region can be described as a i x + b i y + c i ≤ 0, S n is a matrix, each row of which corresponds to the coefficients [a i , b i , c is a vector, each element of which corresponds to c i .

[0113] S334: Formulate the model predictive control problem as a standard quadratic programming problem and solve for the optimal control action through the OSQP solver. The quadratic programming problem is described as:

[0114]

[0115] Where x is the variable to be optimized, Q is the quadratic matrix, c is the linear term, A is the constraint matrix, and l and u are the upper and lower bounds of the constraints respectively.

[0116] Of course, in other embodiments, other solvers can also be selected for solving.

[0117] S4: According to the position judgment of the current legged robot, determine whether the legged robot has always been within the convex region. If not, retreat one prediction step along the historical path and perform path replanning at the same time.

[0118] S5: Convert the control action into an actual control command through differential flatness, and then control the legged robot to navigate.

[0119] Of course, in other embodiments, other methods can also be selected to convert the control action into an actual control command.

[0120] Embodiment 2

[0121] A dynamic response path planning system for a legged robot, comprising:

[0122] A data acquisition module configured to acquire the position information of the legged robot and the perception information of the surrounding environment;

[0123] An obstacle avoidance area construction module, configured to plan a reference path to a target point according to the current position and sensing information, and construct a dynamic obstacle avoidance convex area according to the reference path;

[0124] A path planning and control module, configured to predict the future state of a legged robot based on multiple reference points selected on the reference path, and according to the prediction result, use a model predictive control algorithm to construct and solve an optimization problem including constraints of the dynamic obstacle avoidance convex area, generate an optimal control action, and convert the optimal control action into a control instruction.

[0125] Embodiment III

[0126] A computer-readable storage medium for storing computer instructions, which, when executed by a processor, complete steps S1 - S5 in the method of Embodiment I.

[0127] Embodiment IV

[0128] An electronic device, including a memory, a processor, and computer instructions stored on the memory and running on the processor, which, when run by the processor, complete steps S1 - S5 in the method of Embodiment I.

[0129] Embodiment V

[0130] A legged robot, including a robot body, a controller is provided on the robot body, and computer instructions are stored on the controller to execute steps S1 - S5 in the method of Embodiment I; or a system provided in Embodiment II, or an electronic device provided in Embodiment IV is provided on the robot body.

[0131] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) containing computer-usable program code.

[0132] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate a means for implementing the functions specified in the flow Figure 1 one flow or multiple flows and / or blocks Figure 1 a block or multiple blocks.

[0133] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including an instruction means that implements the functions specified in the flow Figure 1 one flow or multiple flows and / or blocks Figure 1 a block or multiple blocks.

[0134] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in the flow Figure 1 one flow or multiple flows and / or blocks Figure 1 a block or multiple blocks.

[0135] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modifications, equivalent replacements, improvements, etc. made by those skilled in the art without creative efforts within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A dynamic response path planning method for a legged robot, characterized in that: The following steps are involved: Obtain the position information of the legged robot and the perception information of the surrounding environment; According to the current position and perception information, a reference path to the target point is planned, and a dynamic obstacle avoidance convex area is constructed according to the reference path; According to multiple reference points selected on the reference path, the future state of the legged robot is predicted. Based on the prediction results, the model predictive control algorithm is used to construct and solve the optimization problem containing dynamic obstacle avoidance convex area constraints, generate the optimal control action, and convert the optimal control action into control instructions.

2. A dynamic response path planning method for a legged robot as claimed in claim 1, characterized in that: The process of planning a reference path to reach the target point based on the current position and the perception information includes using a heuristic fast path search algorithm to determine the search path based on the map topology of the surrounding environment, and omitting unnecessary nodes during the search process; Or further, creating a first list and a second list, the first list is used to store nodes to be processed that are found during the search process, and the second list is used to store processed nodes; Perform a search, add the current position or starting point of the legged robot to the first list, select the node with the smallest heuristic function from the first list as the current node, and move it to the second list, and when searching, prioritize straight line search according to the direction of the starting point and the target point, then perform an oblique search, and complete the search in the current direction when encountering a jump point or obstacle and add the jump point to the first list; The above search process is repeated until the target point is added to the first list or the first list is empty, and a reference path with the minimum cost value of the cost function is found.

3. A dynamic response path planning method for a legged robot as claimed in claim 1, characterized in that: The process of constructing a dynamic obstacle avoidance convex region according to a reference path includes treating the reference path as a set of discrete points, each discrete point is a path segment, and for each path segment, constructing a rectangle whose center is the midpoint of the line segment, the length of the rectangle is consistent with the direction of the line segment, and the width is perpendicular to the direction of the line segment, and detecting whether the rectangle contains an obstacle. If it contains an obstacle, the width of the rectangle is gradually reduced until the boundary of the rectangle is just tangent to the obstacle; Generate a more accurate convex polygon based on the adjusted rectangle expansion; For each edge of the rectangle, check whether it touches the obstacle. If so, generate a tangent plane in the normal direction. The rectangle is gradually expanded along the path segment direction, and obstacle points are added to generate tangent planes, eventually forming a convex polygon.

4. A dynamic response path planning method for a legged robot as claimed in claim 3, characterized in that when passing through an area whose width is less than a set value or the obstacle distribution density is greater than a set value, the generated convex polygon is retracted, and each edge of the convex polygon is moved inward by a certain length in the normal direction to ensure that the legged robot is always in a safe range. The length of the inward movement is the half diagonal length of the legged robot.

5. The dynamic response path planning method of a legged robot as claimed in claim 1, characterized in that: The process of predicting the future state of the legged robot based on multiple reference points selected on the reference path includes: sampling a series of reference points on the reference path, the positions of each reference point being used as inputs to the model predictive control problem, the first reference position being the point closest to the current position of the legged robot; According to the position of each reference point, the state of the legged robot in the future period of time is obtained based on model predictive control, including position, velocity and acceleration.

6. A dynamic response path planning method for a legged robot as claimed in claim 1, characterized in that: Using the model predictive control algorithm, we construct and solve an optimization problem with dynamic obstacle avoidance convex region constraints. The process of generating the optimal control action includes: The model constraints are described as: X n =f(X n-1 ,U n-1 ) Among them, X follows the initial state of the legged robot, X0 = (p0, v0, a0) T , p, v, a are the current position, velocity and acceleration of the legged robot respectively; The dynamic constraints include upper and lower limits on the velocity and acceleration of the legged robot; The dynamic obstacle avoidance convex area constraint is: S n p n -c≤0; Any line segment in the dynamic obstacle avoidance convex area is described as a i x+b i y+c i ≤0,S n is a matrix whose rows correspond to the coefficients of one side of the polygon [a i ,b i ], c is a vector whose elements correspond to c i ; The model predictive control problem is formulated as a quadratic programming problem and the optimal control action is solved.

7. A dynamic response path planning system for a legged robot, characterized in that: include: A data acquisition module is configured to acquire position information of the legged robot and perception information of the surrounding environment; The obstacle avoidance region construction module is configured to plan a reference path to reach the target point according to the current position and the perception information, and to construct a dynamic obstacle avoidance convex region according to the reference path; The path planning and control module is configured to predict the future state of the legged robot based on multiple reference points selected on the reference path, and based on the prediction results, use the model predictive control algorithm to construct and solve the optimization problem containing dynamic obstacle avoidance convex area constraints, generate optimal control actions, and convert the optimal control actions into control instructions.

8. A computer-readable storage medium, characterized in that: Used to store computer instructions, which, when executed by a processor, complete the steps of the method according to any one of claims 1 to 6.

9. An electronic device, characterized in that: The method comprises a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the steps in the method according to any one of claims 1 to 6 are completed.

10. A leg-foot robot, characterized in that: The robot comprises a robot body, the robot body is provided with a controller, the controller stores computer instructions to execute the steps of the method according to any one of claims 1 to 6; Or the robot body is provided with the system described in claim 7 or the electronic device described in claim 9.

Images