A method for autonomous obstacle avoidance path generation for a robotic arm based on local search algorithm

Abstract

The present invention proposes a method for generating an autonomous obstacle avoidance path for a robotic arm based on a local search algorithm, comprising: step S1, initializing obstacle-related information, and generating a boundary box of the obstacle according to the initialized information; step S2, discretizing the joint points within a neighborhood of the current joint angle; step S3, starting from the current angle, obtaining the next search path point according to the predicted joint angle change law, judging whether a feasible path point can be obtained from the current joint angle, if a feasible path point can be obtained, selecting the optimal path point from the feasible path points, and then judging whether the end point has been reached; if no feasible path point can be obtained, backtracking the path, returning to the previous path point and re-entering S2 to search forward; step S4, if it is judged that the end point has been reached, smoothly connecting all the optimal path points to generate a final smooth path, if it is judged that the end point has not been reached, moving forward one step and returning to step S2 to search for the next path point.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial robots, and in particular to a method for generating an autonomous obstacle avoidance path for a robot arm based on a local search algorithm. Background Art

[0002] In modern robot applications, as the working environment becomes more complex and diverse, the robot arm needs to interact more with objects and humans in the environment, which places higher and higher demands on the robot arm's autonomous obstacle avoidance capability. Good autonomous obstacle avoidance capability can not only greatly increase the robot arm's working efficiency, but also increase the safety of interaction with the human body.

[0003] The existing robotic arm obstacle avoidance solutions are as follows:

[0004] 1. Manual adjustment method. This method requires manually changing the robot's motion path to avoid obstacles. This method often requires multiple attempts to obtain an effective robot motion path. The adjustment process is cumbersome and inefficient, and it is difficult to guarantee the performance of the final path.

[0005] 2. Autonomous planning methods such as artificial potential field method, neural network method, random tree search method, etc. Among them, the artificial potential field method establishes a potential field function between obstacles, target points and robotic arms, and plans the path of the robotic arm by combining the attraction potential of the target point with the repulsion potential of the obstacle. However, this method has the problem of easily falling into local minima and causing wrong planning. The neural network method requires a large amount of data for model training, introduces too much data collection work, and it is difficult to ensure the stability of the algorithm. The random tree search method searches randomly outward from the initial value, continuously expands the search range, and looks for feasible solutions. The algorithm needs to waste a lot of computing power in incorrect search directions, and the algorithm efficiency is not high. Another disadvantage of the above methods is that the shortest path to avoid obstacles is not found, and the curvature of the generated path is not considered, and the final operation efficiency of the robotic arm cannot be guaranteed. Summary of the invention

[0006] The object of the present invention is to solve at least one of the technical drawbacks.

[0007] To this end, the purpose of the present invention is to propose a method for generating an autonomous obstacle avoidance path for a robotic arm based on a local search algorithm.

[0008] In order to achieve the above object, an embodiment of the present invention provides a method for generating an autonomous obstacle avoidance path for a robot arm based on a local search algorithm, comprising the following steps:

[0009] Step S1, initializing obstacle related information, and generating a boundary box of the obstacle according to the initialized information;

[0010] Step S2, discretizing the joint points within a neighborhood of the current joint angle;

[0011] Step S3, starting from the current angle, obtain the next search path point according to the predicted joint angle change law, and determine whether a feasible path point can be obtained from the current joint angle. If a feasible path point can be obtained, select the optimal path point from the feasible path points, and then determine whether the end point has been reached; if no feasible path point can be obtained, backtrack the path, return to the previous path point and re-enter S2 to search forward;

[0012] Step S4: If it is determined that the end point has been reached, all optimal path points are smoothly connected to generate a final smooth path. If it is determined that the end point has not been reached, a step is taken forward and returned to step S2 to search for the next path point.

[0013] Further, in the step S1, the initialization of obstacle-related information includes: initialization of the path starting point, end point, angle limit, and obstacle boundary information.

[0014] Further, the generating of the obstacle bounding box includes: selecting a bounding box of the obstacle described by circular and rectangular elements; wherein the circle is defined by the sphere center O i (xo i ,yo i ,zo i ) and radius ro i The rectangle is composed of the vertices of the cuboid (xc i ,yc i ,zc i ) indicates that the bounding box of the obstacle is described by a cylinder, a sphere, and a polyhedron composed of the above elements.

[0015] Further, in step S2,

[0016] First, each joint angle satisfies the constraints:

[0017]

[0018] in, They represent the joint angle of the i-th joint, the upper limit and lower limit of the joint angle respectively; m represents the total number of joints;

[0019] Formula (1) indicates that each joint angle must be within its limited range, which is an inherent constraint that the robot arm must comply with in any posture;

[0020] According to the specific state of the robot arm, further constraints are imposed on its joint angles:

[0021]

[0022]

[0023]

[0024] in, They represent the angle, upper limit and lower limit of the joint angle of the cth step of the ith joint, respectively, and c represents the step number from the beginning;

[0025] They represent the upper limit of the initial joint angle of the i-th joint, the upper limit of the joint angle of the c-1-th step, and the maximum range of joint angle variation;

[0026] They represent the lower limit of the initial joint angle of the ith joint and the lower limit of the joint angle of the c-1th step respectively;

[0027] Discrete step lengths when determining discrete joint points:

[0028]

[0029] Among them, n s The number of steps when representing discrete joint points, dq i,c represents the discrete step length of the cth step of the i-th joint angle.

[0030] Further, in step S3,

[0031] Step S31, each time the search only selects a discrete point with a discrete step distance from the current point as the target path point, and the possible results for each joint angle in each search are:

[0032]

[0033] in, represents the possible search results of the cth step for the i-th joint angle, q i,c-1 represents the optimal path point of the c-1th step of the i-th joint angle;

[0034] Step S32, for the possible search results in the above formula, determine the search rules and find the optimal path point.

[0035] Further, the determining of the search rules includes:

[0036] Formula (6) can be further expressed as:

[0037]

[0038] Among them, sn i,c is the sign flag, its possible values ​​are -1, 0, 1, where i represents the joint angle number and c represents the step number;

[0039] At this point, the search process of each path point is converted into sn i,c The traversal process uses a method based on predicting the law of joint angle changes to determine sn i,c The traversal criteria enable the optimal path point to be traversed first, thus speeding up the search.

[0040] Furthermore, the traversal rules are as follows:

[0041] In the first step, first take:

[0042]

[0043] in, represents the target angle of the i-th joint angle, represents the initial angle of the i-th joint angle, sgn is the sign function, Indicates sn i,1 The possible values ​​of ;

[0044] According to formula (8), sn is determined i,1 The first value during traversal, the traversal order of the remaining two values ​​is determined by random method, and then sn is determined i,1 The traversal order of

[0045] 2) In the second and third steps, first take:

[0046]

[0047] According to formula (9), sn is determined i,c The first value during traversal is determined by formula (8). If formulas (8) and (9) give the same result, the second value during traversal is determined by random method to determine the traversal order.

[0048] 3) When the number of steps is greater than three, first take:

[0049]

[0050] According to formula (10), sn is determined i,c The first value during traversal; if the first value determined by formula (10) is not 0, the second value during traversal is determined to be 0, and the traversal order is determined; if the first value determined by formula (10) is 0, the second value during traversal is determined by formula (9), if the value determined by formula (9) is also 0, the second value during traversal is determined by formula (8), if the value determined by formula (8) is also 0, the second value during traversal is determined from the remaining two possible values ​​by random method, and finally the traversal order is determined.

[0051] Furthermore, the measurement rules for the quality of discrete path points are determined to screen the discrete path points, and the consumption cost is used to measure the quality of each discrete path point. The specific rules are as follows:

[0052] 1) Determine the current state based on the distance between the current robot arm and the obstacle:

[0053] f obst (q c-1 )>sep1 (11)

[0054] Among them, q c-1 represents the optimal path point of the previous step, the current step is c, f obst It indicates the shortest distance between the robot arm and the obstacle under the input angle, and sep1 indicates the robot arm state threshold;

[0055] When the distance between the robot arm and the obstacle is less than the set threshold, the current state is the surround state, otherwise it is the non-surround state;

[0056] Determine the relationship between the searched path points and obstacles:

[0057]

[0058] in, It indicates the currently searched path point, that is, the set of joint angles of the current path point. sep2 indicates the minimum distance allowed for the robot rod to be away from obstacles.

[0059] If the minimum distance between the searched path point and the obstacle is greater than the set threshold, the current path point is marked as a feasible path point and enters the subsequent screening. Otherwise, the point is directly eliminated and the next discrete path point is screened.

[0060] If the current state is a non-circular state, the consumption cost of each feasible path point is determined by the shortest distance principle:

[0061]

[0062] Among them, cost represents the consumption cost under the input angle; the above formula takes the distance between the search path point and the final target point as the consumption cost of the current path point.

[0063] 4) If the current state is a wrap-around state, the consumption component of each feasible path point is determined by the shortest distance principle and the best obstacle avoidance principle:

[0064]

[0065] Among them, k1 represents the weight coefficient of the shortest path criterion, and k2 represents the weight coefficient of the obstacle distance principle.

[0066] Furthermore, the path backtracking includes the following steps:

[0067] When from q c To q c+1 During the search, if it is found that no feasible path point can be obtained, the current path state is changed from q c Back to q c-1 state, exclude the optimal point found in the last planning, select the optimal point from the remaining points, and repeat the process of moving to q c search.

[0068] Further, in step S4,

[0069] Use quintic spline curve to generate the final path:

[0070] q i =a i l 5 +b i l 4 +c i l 3 +d i l 2 +e i l+f i (15)

[0071] Among them, a i ,b i ,c i ,d i ,e i ,f i represents the parameters of the quintic spline curve describing the i-th joint angle, and l is the formal parameter; taking the first joint angle q1 as the formal parameter, it is convenient to determine the relationship between the joint angle and the formal parameter at each point; at this time, the quintic spline curve describing the joint angle q1 degenerates into a one-dimensional straight line, and the remaining joint angles are normal quintic spline curves; at this point, a complete obstacle avoidance path curve of the robotic arm is generated.

[0072] The method for generating an autonomous obstacle avoidance path for a robotic arm based on a local search algorithm according to an embodiment of the present invention has the following beneficial effects:

[0073] 1. Automatically search for feasible trajectories, eliminating the tedious manual adjustment of paths;

[0074] 2. Dynamically adjust the search step size based on the position relationship with the obstacle, reducing the amount of calculation and increasing the smoothness of the trajectory;

[0075] 3. By predicting and adjusting the search rules at each step, the search speed is accelerated and the amount of calculation is reduced;

[0076] 4. The state of each step is divided into a surrounding state and a non-surrounding state to ensure the smoothness of the generated trajectory.

[0077] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0079] Figure 1 A flowchart of a method for generating an autonomous obstacle avoidance path for a robotic arm based on a local search algorithm according to an embodiment of the present invention;

[0080] Figure 2 Schematic diagram of a method for generating an autonomous obstacle avoidance path for a robotic arm based on a local search algorithm according to an embodiment of the present invention. DETAILED DESCRIPTION

[0081] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

[0082] An embodiment of the present invention provides a method for generating an autonomous obstacle avoidance path for a robotic arm based on a local search algorithm. The method adopts a local search algorithm, takes into account the length and curvature of the path, searches from the initial angle of the robotic arm, searches for the optimal path point for the next step from the current path point in each step, and then smoothly connects all the optimal path points to generate an optimal path that meets the conditions.

[0083] like Figure 1 and Figure 2 As shown, the method for generating an autonomous obstacle avoidance path for a manipulator based on a local search algorithm according to an embodiment of the present invention comprises the following steps:

[0084] Step S1, initializing obstacle related information, and generating a boundary box of the obstacle according to the initialized information.

[0085] Initialize obstacle-related information, including: initialization path starting point, end point, angle limit, and obstacle boundary information.

[0086] Specifically, generating a bounding box of an obstacle includes: selecting a bounding box of the obstacle described by circular and rectangular elements; wherein the circle is defined by the center Oi(xo i ,yo i ,zo i ) and radius roi The rectangle is composed of the vertices of the cuboid (xc i ,yc i ,zc i ) indicates that the bounding box of the obstacle is described by a cylinder, a sphere, and a polyhedron composed of the above elements.

[0087] Step S2, discretize the joint points within a neighborhood of the current joint angle.

[0088] First, each joint angle satisfies the constraints:

[0089]

[0090] in, They represent the joint angle of the i-th joint, the upper limit and lower limit of the joint angle respectively; m represents the total number of joints.

[0091] Formula (1) indicates that each joint angle must be within its limited range, which is an inherent constraint that the robot arm must comply with in any posture.

[0092] In the method of the present invention, further constraints are imposed on the joint angles of the robot according to the specific state of the robot:

[0093]

[0094]

[0095]

[0096] in, They represent the angle, upper limit and lower limit of the joint angle of the cth step of the ith joint, respectively, and c represents the step number from the beginning;

[0097] They respectively represent the upper limit of the initial joint angle of the ith joint, the upper limit of the joint angle of the c-1th step, and the maximum range of joint angle changes.

[0098] They represent the lower limit of the initial joint angle of the ith joint and the lower limit of the joint angle of the c-1th step respectively.

[0099] On this basis, the discrete step lengths at discrete joint points are determined:

[0100]

[0101] Among them, n s The number of steps when representing discrete joint points, dq i,c represents the discrete step length of the cth step of the i-th joint angle.

[0102] Step S3, starting from the current angle, obtain the next search path point based on the predicted joint angle change law, and determine whether a feasible path point can be obtained from the current joint angle. If a feasible path point can be obtained, select the optimal path point from the feasible path points, and then determine whether the end point has been reached; if a feasible path point cannot be obtained, backtrack the path, return to the previous path point and re-enter S2 to search forward.

[0103] Specifically, the optimal path point search module is responsible for selecting the optimal path point that meets the conditions from the discrete points starting from the initial angle, continuously searching and advancing, and finally reaching the target point. The method of the present invention adopts a method based on the law of predicting the change of joint angles to quickly search for discrete path points and search for the optimal path point.

[0104] First, in order to reduce the amount of calculation and speed up the path search, each search only selects a discrete point with a discrete step length away from the current point as the target path point. Therefore, the possible results of each joint angle in each search are:

[0105]

[0106] in, represents the possible search results of the cth step for the i-th joint angle, q i,c-1 represents the optimal path point of the c-1th step for the ith joint angle.

[0107] Secondly, for the three possible search results shown in equation (6), it is necessary to determine the search rules so as to find the optimal path point as quickly as possible.

[0108] The search rules are as follows:

[0109] Formula (6) can be further expressed as:

[0110]

[0111] Among them, sn i,c It is the sign flag, and its possible values ​​are -1, 0, 1, where i represents the joint angle number and c represents the step number.

[0112] At this point, the search process of each path point is converted into sn i,c The method of the present invention uses a method based on predicting the law of joint angle change to determine sn i,c The traversal rule enables the optimal path point to be traversed first, thus speeding up the search. The traversal rules are as follows:

[0113] 1) In the first step, first take:

[0114]

[0115] in, represents the target angle of the i-th joint angle, represents the initial angle of the i-th joint angle, sgn is a function of positive and negative signs, Indicates sn i,1 The possible values ​​of .

[0116] According to formula (8), sn is determined i,1 The first value during traversal, the traversal order of the remaining two values ​​is determined by random method, and then sn is determined i,1 The traversal order.

[0117] 2) In the second and third steps, first take:

[0118]

[0119] Similarly, sn is determined according to equation (9): i,c The first value during traversal is determined by formula (8). If formulas (8) and (9) give the same result, the second value during traversal is determined by random method to determine the traversal order.

[0120] 3) When the number of steps is greater than three, first take:

[0121]

[0122] Similarly, sn is determined according to equation (10): i,c The first value during traversal. If the first value determined by formula (10) is not 0, the second value during traversal is determined to be 0, and the traversal order is determined. If the first value determined by formula (10) is 0, the second value during traversal is determined by formula (9). If the value determined by formula (9) is also 0, the second value during traversal is determined by formula (8). If the value determined by formula (8) is also 0, the second value during traversal is determined from the remaining two possible values ​​using a random method, and the traversal order is finally determined.

[0123] The above search rules predict the changing direction of the trajectory and ensure that the optimal path point is searched at the fastest speed.

[0124] On this basis, the discrete path points can be screened by determining the measurement rules for the advantages and disadvantages of the discrete path points.

[0125] The method of the present invention uses consumption cost to measure the quality of each discrete path point. The specific rules are as follows:

[0126] 1) Determine the current state based on the distance between the current robot arm and the obstacle:

[0127] f obst (q c-1)>sep1 (11)

[0128] Among them, q c-1 represents the optimal path point of the previous step (currently step c), f obst It indicates the shortest distance between the robot arm and the obstacle under the input angle, and sep1 indicates the robot arm state threshold.

[0129] When the distance between the robot arm and the obstacle is less than the set threshold, the current state is the surround state, otherwise it is the non-surround state.

[0130] 2) Determine the relationship between the searched path points and obstacles:

[0131]

[0132] in, It represents the currently searched path point (i.e., the set of joint angles of the current path point), and sep2 represents the minimum distance allowed for the robot rod to be away from obstacles.

[0133] If the minimum distance between the searched path point and the obstacle is greater than the set threshold, the current path point is marked as a feasible path point and enters the subsequent screening. Otherwise, the point is directly eliminated and the next discrete path point is screened.

[0134] 3) If the current state is a non-circular state, the consumption cost of each feasible path point is determined by the shortest distance principle:

[0135]

[0136] Among them, cost represents the consumption cost under the input angle.

[0137] The above formula uses the distance between the search path point and the final target point as the consumption cost of the current path point.

[0138] 4) If the current state is a wrap-around state, the consumption component of each feasible path point is determined by the shortest distance principle and the best obstacle avoidance principle:

[0139]

[0140] Among them, k1 represents the weight coefficient of the shortest path criterion,

[0141] k2 represents the weight coefficient of the obstacle distance principle.

[0142] This method will not cause the robot arm to be too close to the obstacle due to searching for the shortest path, nor will it cause the path to be too long due to searching for the best obstacle avoidance path.

[0143] In the above step length setting algorithm, the shortest path principle reduces the length of the path and improves the working efficiency of the robot. By setting a reasonable surround area threshold sep1 and the weight coefficient k2 of the obstacle distance principle, it can be ensured that when the robot bypasses an obstacle, the curvature of the path is kept at a low level, thereby improving the operating speed of the robot.

[0144] When the above algorithm is used for searching, if the convergence condition is met, or if the current point is directly determined to be the shortest path point in the non-circular state, the current search point can be set as the final path point. When it is difficult to determine the final path point, the search point with the minimum cost is selected from all the search points that meet the conditions as the final path point.

[0145] cost(q c )<=sep cost (15)

[0146] Among them, sep cost Indicates the cost consumption threshold.

[0147] In order to avoid the search algorithm failing to find the global optimal value due to the influence of a local optimal value, path backtracking is used to solve this problem. The path backtracking process includes the following steps:

[0148] When from q c To q c+1 During the search, if it is found that no feasible path point can be obtained, the current path state is changed from q c Back to q c-1 state, exclude the optimal point found in the last planning, select the optimal point from the remaining points, and repeat the process of moving to q c In order to save computational effort, a suboptimal point register needs to be added to the optimal path point search module to save the suboptimal points encountered during the search of the last two nodes so that they can be used directly when tracing back the path.

[0149] Step S4: If it is determined that the end point has been reached, all optimal path points are smoothly connected to generate a final smooth path. If it is determined that the end point has not been reached, a step is taken forward and returned to step S2 to search for the next path point.

[0150] The method of the present invention uses a quintic spline curve to generate the final path:

[0151] q i =a i l 5 +b i l 4 +c i l 3 +d i l 2 +e i l+fi (16)

[0152] Among them, a i ,b i ,c i ,d i ,e i ,f i Represents the parameters of the quintic spline curve describing the i-th joint angle, and l is a formal parameter.

[0153] Here, the first joint angle q1 is taken as the formal parameter to facilitate the determination of the relationship between the joint angle and the formal parameter at each point.

[0154] At this point, the quintic spline curve describing the joint angle q1 degenerates into a one-dimensional straight line, and the remaining joint angles are normal quintic spline curves. At this point, a complete obstacle avoidance path curve for the robotic arm is generated.

[0155] The method for generating an autonomous obstacle avoidance path for a robotic arm based on a local search algorithm according to an embodiment of the present invention has the following beneficial effects:

[0156] 1. Automatically search for feasible trajectories, eliminating the tedious manual adjustment of paths;

[0157] 2. Dynamically adjust the search step size based on the position relationship with the obstacle, reducing the amount of calculation and increasing the smoothness of the trajectory;

[0158] 3. By predicting and adjusting the search rules at each step, the search speed is accelerated and the amount of calculation is reduced;

[0159] 4. The state of each step is divided into a surrounding state and a non-surrounding state to ensure the smoothness of the generated trajectory.

[0160] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0161] Although the embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and cannot be understood as limiting the present invention. Those skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention without departing from the principles and purpose of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for generating an autonomous obstacle avoidance path for a robotic arm based on a local search algorithm, characterized in that: The steps include: Step S1, initializing obstacle related information, and generating a boundary box of the obstacle according to the initialized information; Step S2, discretizing the joint points within a neighborhood of the current joint angle; Step S3, starting from the current angle, obtaining the next search path point according to the predicted joint angle change law; judging whether a feasible path point can be obtained from the current joint angle, and if a feasible path point can be obtained, selecting the optimal path point from the feasible path points, and then judging whether the end point has been reached; If no feasible path point can be obtained, the path is backtracked, returning to the previous path point and re-entering S2 to search forward; In step S3, Step S31, each time the search only selects a discrete point with a discrete step distance from the current point as the target path point, and the possible results for each joint angle in each search are: in, represents the possible search results of the cth step for the i-th joint angle, q i,c-1 represents the optimal path point of the c-1th step of the i-th joint angle; Step S32, for the possible search results in the above formula, determine the search rules to find the optimal path point; wherein the determining the search rules includes: Formula (6) can be further expressed as: Among them, sn i,c is the sign flag, its possible values ​​are -1, 0, 1, where i represents the joint angle number and c represents the step number; At this point, the search process of each path point is converted into sn i,c The traversal process uses a method based on predicting the law of joint angle changes to determine sn i,c The traversal criteria enable the optimal path point to be traversed first, thus speeding up the search; The traversal criteria are as follows: 1) In the first step, first take: in, represents the target angle of the i-th joint angle, represents the initial angle of the i-th joint angle, sgn is the sign function, Indicates sn i,1 The possible values ​​of ; According to formula (8), sn is determined i,1 The first value during traversal, the traversal order of the remaining two values ​​is determined by random method, and then sn is determined i,1 The traversal order of 2) In the second and third steps, first take: According to formula (9), sn is determined i,c The first value during traversal is determined by formula (8). If formulas (8) and (9) give the same result, the second value during traversal is determined by random method to determine the traversal order. 3) When the number of steps is greater than three, first take: According to formula (10), sn is determined i,c The first value during traversal; if the first value determined by formula (10) is not 0, the second value during traversal is determined to be 0, and the traversal order is determined; if the first value determined by formula (10) is 0, the second value during traversal is determined by formula (9), if the value determined by formula (9) is also 0, the second value during traversal is determined by formula (8), if the value determined by formula (8) is also 0, the second value during traversal is determined from the remaining two possible values ​​by random method, and finally the traversal order is determined; Step S4: If it is determined that the end point has been reached, all optimal path points are smoothly connected to generate a final smooth path. If it is determined that the end point has not been reached, a step is taken forward and returned to step S2 to search for the next path point.

2. The method for generating an autonomous obstacle avoidance path for a robotic arm based on a local search algorithm according to claim 1, characterized in that: In the step S1, the initialization of obstacle-related information includes: initialization of the path starting point, end point, angle limit, and obstacle boundary information.

3. The method for generating an autonomous obstacle avoidance path for a robotic arm based on a local search algorithm according to claim 1, characterized in that: The generating of the obstacle bounding box comprises: selecting a bounding box of the obstacle described by circular and rectangular elements; wherein the circle is defined by the sphere center O i (xo i ,yo i ,zo i ) and radius ro i The rectangle is composed of the vertices of the cuboid (xc i ,yc i ,zc i ) indicates that the bounding box of the obstacle is described by a cylinder, a sphere, and a polyhedron composed of the above elements.

4. The method for generating an autonomous obstacle avoidance path for a robotic arm based on a local search algorithm according to claim 1, characterized in that: In step S2, First, each joint angle satisfies the constraints: Among them, q i , They represent the joint angle of the i-th joint, the upper limit and lower limit of the joint angle respectively; m represents the total number of joints; Formula (1) indicates that each joint angle must be within its limited range, which is an inherent constraint that the robot arm must comply with in any posture; According to the specific state of the robot arm, further constraints are imposed on its joint angles: Among them, q i,c , They represent the angle, upper limit and lower limit of the joint angle of the cth step of the ith joint, respectively, and c represents the step number from the beginning; r i,c They represent the upper limit of the initial joint angle of the i-th joint, the upper limit of the joint angle of the c-1-th step, and the limit change range of the joint angle respectively; They represent the lower limit of the initial joint angle of the ith joint and the lower limit of the joint angle of the c-1th step respectively; Discrete step lengths when determining discrete joint points: Among them, n s The number of steps when representing discrete joint points, dq i,c represents the discrete step length of the cth step of the i-th joint angle.

5. The method for generating an autonomous obstacle avoidance path for a robotic arm based on a local search algorithm according to claim 1, characterized in that: Determine the measurement rules for the quality of discrete path points. The discrete path points are screened and the consumption cost is used to measure the quality of each discrete path point. The specific rules are as follows: 1) Determine the current state based on the distance between the current robot arm and the obstacle: f obst (what c-1 )>sep1(11) Among them, q c-1 represents the optimal path point of the previous step, the current step is c, f obst It indicates the shortest distance between the robot arm and the obstacle under the input angle, and sep1 indicates the robot arm state threshold; When the distance between the robot arm and the obstacle is less than the set threshold, the current state is the surround state, otherwise it is the non-surround state; Determine the relationship between the searched path points and obstacles: in, It indicates the currently searched path point, that is, the set of joint angles of the current path point. sep2 indicates the minimum distance allowed for the robot rod to be away from obstacles. If the minimum distance between the searched path point and the obstacle is greater than the set threshold, the current path point is marked as a feasible path point and enters the subsequent screening. Otherwise, the point is directly eliminated and the next discrete path point is screened. If the current state is a non-circular state, the consumption cost of each feasible path point is determined by the shortest distance principle: Among them, cost represents the consumption cost under the input angle; the above formula takes the distance between the search path point and the final target point as the consumption cost of the current path point; 4) If the current state is a wraparound state, the consumption cost of each feasible path point is determined by the shortest distance principle and the best obstacle avoidance principle: Among them, k1 represents the weight coefficient of the shortest path criterion, and k2 represents the weight coefficient of the obstacle distance principle.

6. The method for generating an autonomous obstacle avoidance path for a robotic arm based on a local search algorithm according to claim 1, characterized in that: The path backtracking includes the following steps: When from q c To q c+1 During the search, if it is found that no feasible path point can be obtained, the current path state is changed from q c Back to q c-1 state, exclude the optimal point found in the last planning, select the optimal point from the remaining points, and re-plan to q c search.

7. The method for generating an autonomous obstacle avoidance path for a robotic arm based on a local search algorithm according to claim 1, characterized in that: In step S4, Use quintic spline curve to generate the final path: q i =a i l 5 +b i l 4 +c i l 3 +d i l 2 +e i l+f i (15) Among them, a i ,b i ,c i ,d i ,e i ,f i represents the parameters of the quintic spline curve describing the i-th joint angle, and l is the formal parameter. Taking the first joint angle q1 as the formal parameter makes it easier to determine the relationship between the joint angle and the formal parameter at each point. At this point, the quintic spline curve describing the joint angle q1 degenerates into a one-dimensional straight line, and the remaining joint angles are normal quintic spline curves. At this point, a complete obstacle avoidance path curve for the robotic arm is generated.

Images