Mechanical arm trajectory planning method based on structure-oriented motion profile

Abstract

A mechanical arm trajectory planning method based on a structure-oriented motion profile belongs to the related technical field of robot control, and comprises the following steps: S1, inputting an expected position point matrix P of a joint space and physical constraints of a mechanical arm; s2, initializing parameters, wherein the parameters comprise a slope coefficient and a local asymmetric coefficient of the point-to-point trajectory model, a variable boundary, a population size, an external memory size and a maximum number of iterations; s3, generating an initial track Omega based on the structure-oriented motion profile; s4, optimizing an initial track omega according to an analytical method priority strategy; and S5, outputting an optimal solution or a Pareto optimal solution set. Efficient and high-precision operation of the mechanical arm is achieved, and the production efficiency and the operation precision are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field related to robot control, and in particular relates to a robot arm trajectory planning method based on structure-guided motion profile. Background Art

[0002] In modern automation scenarios, high-precision robotic arms are the best substitute for human resources. Through precise control and flexible movement, they can perform repetitive and complex tasks such as gluing, assembly, and welding, which can significantly improve production efficiency and reduce the risk of workers working in high-temperature and high-risk environments. The robotic arm is a complex mechanical mechanism composed of joints, connecting rods, drives, sensors, and control systems. Different numbers and types of joint degrees of freedom can be designed according to the usage scenario. Trajectory planning is a basic technology for robotic arms, which determines the smoothness and accuracy of joint motion and end effector motion. Compared with Cartesian space, trajectory planning in joint space can avoid complex kinematic singularity calculations and can directly achieve high-precision control of joint drives. In addition, the smoothness of the trajectory plays an important role in improving the life of the robotic arm and reducing errors caused by joint vibration.

[0003] With the continuous deepening of research on trajectory planning technology, a large number of studies at home and abroad have used spline functions such as polynomials, cubic splines, and quintic B-splines to interpolate trajectories, and used optimization algorithms such as particle swarms, genetic algorithms, and whale algorithms to find the optimal trajectory. However, since there is no velocity constraint on the desired position point, most of the existing methods are difficult to avoid overshoot and undershoot in the trajectory. In addition, there is currently no method that can integrate multi-point trajectory models and point-to-point trajectory models for trajectory interpolation, and integrate analytical and numerical methods to calculate the optimal trajectory. These areas that need to be improved may have a negative impact on the tracking accuracy of the end effector of the robot arm, reduce the operating accuracy of the robot arm, and limit the application of the robot arm in high-precision operation scenarios. Summary of the invention

[0004] In order to overcome the shortcomings of the existing technology, the present invention provides a robot arm trajectory planning method based on structure-guided motion profile and analytical method priority strategy, so as to realize efficient and high-precision operation of the robot arm and improve production efficiency and operation accuracy.

[0005] The technical solution adopted by the present invention to solve its technical problem is:

[0006] A robot arm trajectory planning method based on structure-guided motion profile comprises the following steps:

[0007] S1, input the desired position point matrix P of the joint space and the physical constraints of the robot;

[0008] S2, initializing parameters, the parameters including: slope coefficient and local asymmetric coefficient of the point-to-point trajectory model, variable boundary, population size, external memory size and maximum number of iterations;

[0009] S3, generating an initial trajectory Ω based on the structure-guided motion profile;

[0010] S4, optimize the initial trajectory Ω according to the analytical priority strategy;

[0011] S5. Output the optimal solution or Pareto optimal solution set.

[0012] Furthermore, in S1, the expected position points of the input joint space are an n×m matrix, where n represents the number of joints and m represents the number of expected positions; the physical constraints of the robotic arm are set according to user needs, including joint angle constraints, joint velocity constraints, joint acceleration constraints and joint velocity and acceleration constraints.

[0013] Furthermore, in S3, the process of generating the initial trajectory Ω based on the structure-guided motion profile is:

[0014] S31, generating a speed constraint attribute list corresponding to the expected position point matrix;

[0015] S32, according to the generated speed constraint attribute list, splitting the expected position point matrix P into a plurality of multi-point expected position point matrices and a plurality of point-to-point expected position point matrices according to the joints at which they are located;

[0016] S33, the multi-point expected position matrix and the point-to-point expected position matrix generate corresponding trajectories using the multi-point trajectory model and the point-to-point trajectory model respectively;

[0017] S34. According to the position of the desired position point in the matrix, the obtained multi-point trajectory and the point-to-point trajectory are combined to form a complete trajectory Ω corresponding to the desired position point matrix P.

[0018] Preferably, in said S31, the generated velocity constraint attribute list is an n×m matrix, wherein n represents the number of joints, and m represents the number of expected joint positions;

[0019] Each value vc in the velocity constraint attribute list is determined by the movement direction of two adjacent sub-trajectories, and the calculation formula is:

[0020]

[0021] Among them, p i,g is the expected position of the i-th joint of the g-th joint, 1 represents the forward motion of the joint, -1 represents the reverse motion of the joint, and 0 represents the stop motion of the joint.

[0022] More preferably, in S32, the multi-point expected position matrix is ​​a one-dimensional matrix, the number of joint expected position points exceeds 2, and the velocity constraint attribute values ​​of the start and end joint expected position points are both equal to 0;

[0023] The point-to-point expected position matrix is ​​a one-dimensional matrix. The number of joint expected position points is equal to 2, and the velocity constraint attribute values ​​of the two joint expected position points are both equal to 0.

[0024] More preferably, in said S33, any multi-point trajectory model and point-to-point trajectory model may be used to generate the corresponding trajectory, and after disassembly and reassembly, the complete trajectory model of each joint may be different;

[0025] The point-to-point desired position matrix generates the corresponding trajectory using the local asymmetric acceleration motion profile, where the calculation formula for the local asymmetric acceleration motion profile is:

[0026]

[0027] Where J(t) is the time-varying acceleration, j peak is the peak jerk during joint motion, t i is the end time of the ith trajectory segment, h i is the running time of the i-th trajectory segment, t is the time variable;

[0028] The displacement, velocity, and acceleration of the point-to-point trajectory are obtained by integrating the jerk;

[0029] The multi-point expected position matrix uses the B-spline function to generate the corresponding trajectory, where the displacement θ is expressed by the quintic B-spline function as follows:

[0030]

[0031] Where: cp ni is the ni-th control point of the quintic B-spline function, u is the normalized knot vector, n cp is the total number of control points, N ni,5 (u) is the ni-th basis function;

[0032] N ni,5 (u) is obtained using the de Boer-Cox recursion formula,

[0033]

[0034] Among them, u ni represents the nith node in the normalized node vector variable, N ni,j (u) is the ni-th j-th basis function, m1 and m2 are the two coefficients of the basis function;

[0035] According to the position of the desired position point in the matrix, the obtained multi-point trajectory and point-to-point trajectory are combined to form a complete trajectory Ω corresponding to the desired position point matrix P;

[0036] Structure-guided motion profiles can generate different complete trajectory models for different joints based on a list of velocity constraint attributes.

[0037] In S4, the point-to-point trajectory model is solved by an analytical method, and a unique and feasible solution can be obtained. The process of optimizing the initial trajectory Ω according to the analytical method priority strategy is:

[0038] S41, evaluating and updating the lower bound of the running time of each sub-trajectory according to the point-to-point trajectory model used;

[0039] S42, optimizing the trajectory Ω using the obtained lower bound of the running time of each sub-trajectory, and determining whether the optimized trajectory Ω meets the target condition;

[0040] S43, if the target conditions are met, the optimal solution is output and the trajectory planning is completed;

[0041] S44. If the target conditions are not met, the numerical calculation phase is entered until the termination conditions are met, and the optimal solution or Pareto optimal solution set is output to complete the trajectory planning.

[0042] Preferably, in S41, the variable lower bounds set for the running time of all sub-trajectories are assigned, and for the point-to-point sub-trajectory, the running time of the sub-trajectory is solved, and finally the running time of the i-th sub-trajectory of all joints is determined according to the following formula:

[0043]

[0044] Among them, et i represents the running time of the ith sub-trajectory, Represents the running time of the i-th point-to-point trajectory type sub-trajectory of the g-th joint.

[0045] More preferably, in said S42, the objective function includes trajectory running time, overshoot and undershoot, and energy consumption and driving torque can be increased according to user needs; any multi-objective optimization algorithm and constraint control method are used to optimize the trajectory;

[0046] Target conditions include:

[0047] C1: at least one joint of the ith sub-trajectory is of point-to-point type;

[0048] C2: The generated complete trajectory does not have overshoot and undershoot;

[0049] C3: The generated complete trajectory satisfies the physical constraints of the robot.

[0050] More preferably, in S44, the termination condition mainly refers to that the number of iterations has reached the maximum iter_max; if the Pareto optimal solution set is obtained, the user selects the optimal solution according to the comprehensive fuzzy membership function of the Pareto frontier to generate the optimized trajectory, and the process is as follows:

[0051] The multi-objective improved student psychology optimization algorithm and adaptive ε constraint technology are used for optimization. The population is divided into four categories of individuals: the best student, good student, ordinary student and students to be improved; the best student is the best individual in the population, and the selection of good students, ordinary students and students to be improved is random;

[0052] The best students' scores were improved using the following formula,

[0053]

[0054] in, is the best student individual at the iterth iteration, is the random student individual at the iter iteration, rand is a random number in [0,1], r 0 is a random number of 1 or 2;

[0055] The scores of good students are improved using the following two formulas, and the selection is also random:

[0056]

[0057] in, is the individual student whose score is the average, rand1 is a random number in [0,1], and rand2 is a random number in [0,1];

[0058] The score improvement of ordinary students adopts the following formula,

[0059]

[0060] The scores of students to be improved are improved using the following two formulas, and the selection is also random:

[0061]

[0062]

[0063] Where, L(λ) is the Levy flight function;

[0064] The update of ε in the adaptive ε constraint technique is related to the number of feasible solutions and the current iteration number, as shown below:

[0065]

[0066] Among them, c fis the ratio of feasible solutions to all solutions in the current iteration, c iter It is the ratio of the current cumulative number of iterations to the maximum number of iterations, and iter_max is the maximum number of iterations;

[0067] The fast non-dominated sorting method is used to sort the individuals in the population, and the non-dominated individuals are selected and stored in the external memory;

[0068] When a new individual needs to be stored in the external memory, if the new individual is dominated by at least one individual in the external memory, then the new individual is not allowed to be placed in the external memory; if the new individual dominates existing individuals in the external memory, then these existing individuals are removed and the new individual is stored; if the new individual does not dominate existing individuals in the external memory and the existing individuals in the external memory do not dominate the new individual, then the new individual is stored in the external memory; if the new individual is allowed to be stored in the external memory and the number of individuals in the external memory has reached the upper limit, then individuals with low density are removed until the number of individuals in the external memory is equal to the upper limit;

[0069] The density of individuals is estimated using the following formula,

[0070]

[0071] Among them, D x (id) is the density of the idth individual in the decision space, D f (id) is the density of the idth element in the target space, and are the maximum and minimum values ​​of the j-th dimension variable, respectively. and are the maximum and minimum values ​​of the objective function of the jth dimension, respectively, and class_size is the population size;

[0072] Determine whether the termination condition is met. If not, continue to execute the multi-objective improved student psychology optimization algorithm; if it is met, terminate the calculation and output the Pareto optimal solution set;

[0073] The user selects the optimal solution based on the comprehensive fuzzy membership function of the Pareto front to generate the optimized trajectory.

[0074] s syn (id)=(s1(id)+s2(id)) / max{s1+s2};

[0075]

[0076] Among them, f j (id) is the jth objective function value corresponding to the idth optimal solution, s j (id) is the fuzzy membership function of the jth objective of the idth optimal solution, ssyn (id) is the comprehensive membership value of the idth optimal solution.

[0077] The beneficial effects of the present invention are mainly manifested in:

[0078] 1. Speed ​​constraints are imposed at the desired position points, providing an integrated framework of multi-point trajectory models and point-to-point trajectory models, reducing or even avoiding overshoot and undershoot in the trajectory, and improving the operation accuracy of the robot arm;

[0079] 2. It integrates analytical and numerical methods, and gives priority to analytical methods, which can quickly generate the optimal solution or provide a Pareto optimal solution set for users to choose according to actual needs, thereby improving the operating efficiency and accuracy of the robotic arm while meeting the diverse needs of users. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] Figure 1 It is a schematic diagram of the overall method flow of this application;

[0081] Figure 2 It is a schematic diagram of generating an initial trajectory based on a structure-guided motion profile of the present application;

[0082] Figure 3 It is the optimal jerk curve of time-overshoot-undershoot of each joint of a six-degree-of-freedom robot arm for working condition 1 of the present application.

[0083] Figure 4 It is the Pareto optimal solution set of a certain six-degree-of-freedom robotic arm of this application for working condition 2.

[0084] Figure 5 It is the optimal acceleration curve of each joint time of a six-degree-of-freedom robot arm for working condition 2 of the present application.

[0085] Figure 6 It is the optimal acceleration curve of overshoot and undershoot of each joint of a six-degree-of-freedom robot arm for working condition 2 of the present application. DETAILED DESCRIPTION

[0086] The present invention will be further described below in conjunction with the accompanying drawings.

[0087] Reference Figure 1 to Figure 6 , a robot arm trajectory planning method based on structure-guided motion profile, comprising the following steps:

[0088] S1. Input the desired position point matrix P of the joint space and the physical constraints of the robot.

[0089] For working condition 1, the expected position data of each joint and the physical constraints of robot 1 are shown in Table 1 below:

[0090]

[0091]

[0092] Table 1

[0093] For working condition 2, the expected position data of each joint and the physical constraints of robot 2 are shown in Table 2 below:

[0094]

[0095] Table 2

[0096] S2, initialization parameters, including general parameters and specific parameters. General parameters are generally irrelevant to the performance of the planning method, and mainly include variable boundaries, population size class_size, external memory size archive_size, and maximum number of iterations iteration_max, etc. Specific parameters are related to the performance of the planning method and depend on the trajectory model, optimization algorithm, and constraint processing method used;

[0097] The parameters of this embodiment include: a slope coefficient and a local asymmetric coefficient of a point-to-point trajectory model, a variable boundary, a population size, an external memory size, and a maximum number of iterations.

[0098] S3, generating the initial trajectory Ω based on the structure-guided motion profile, the process is as follows:

[0099] S31, generating a speed constraint attribute list corresponding to the expected position point matrix;

[0100] The values ​​in the velocity constraint attribute list are mainly determined by the movement directions of two adjacent sub-trajectories. The calculation formula is:

[0101]

[0102] Among them, p i,g is the expected position of the i-th joint of the g-th joint, 1 represents the forward motion of the joint, -1 represents the reverse motion of the joint, and 0 represents the stop motion of the joint;

[0103] S32, according to the generated speed constraint attribute list, splitting the expected position point matrix P into a plurality of multi-point expected position point matrices and a plurality of point-to-point expected position point matrices according to the joints at which they are located;

[0104] The multi-point expected position matrix is ​​a one-dimensional matrix. The number of joint expected position points exceeds 2, and the velocity constraint attribute values ​​of the start and end joint expected position points are both equal to 0.

[0105] The point-to-point expected position matrix is ​​a one-dimensional matrix. The number of joint expected position points is equal to 2, and the velocity constraint attribute values ​​of the two joint expected position points are both equal to 0.

[0106] S33, the multi-point expected position matrix and the point-to-point expected position matrix generate corresponding trajectories using the quintic B-spline function and the local asymmetric acceleration motion profile respectively;

[0107] Any multi-point trajectory model and point-to-point trajectory model can be used to generate the corresponding trajectory. After disassembly and reassembly, the complete trajectory model of each joint may be different;

[0108] The point-to-point desired position matrix generates the corresponding trajectory using the local asymmetric acceleration motion profile, where the calculation formula for the local asymmetric acceleration motion profile is:

[0109]

[0110] Where J(t) is the time-varying acceleration, j peak is the peak jerk during joint movement, t i is the end time of the ith trajectory segment, h i is the running time of the i-th trajectory segment, t is the time variable;

[0111] The displacement, velocity, and acceleration of the point-to-point trajectory are obtained by integrating the jerk;

[0112] The multi-point expected position matrix uses the B-spline function to generate the corresponding trajectory, where the displacement θ is expressed by the quintic B-spline function as follows:

[0113]

[0114] Where: cp ni is the ni-th control point of the quintic B-spline function, u is the normalized knot vector, n cp is the total number of control points, N ni,5 (u) is the ni-th basis function;

[0115] N ni,5 (u) is obtained using the de Boer-Cox recursion formula,

[0116]

[0117] Among them, u ni represents the nith node in the normalized node vector variable, N ni,j (u) is the ni-th j-th basis function, m1 and m2 are the two coefficients of the basis function;

[0118] According to the position of the desired position point in the matrix, the obtained multi-point trajectory and point-to-point trajectory are combined to form a complete trajectory Ω corresponding to the desired position point matrix P;

[0119] The structure-guided motion profile can generate different complete trajectory models for different joints according to the velocity constraint attribute list;

[0120] S34, according to the position of the desired position point in the matrix, the obtained multi-point trajectory and the point-to-point trajectory are combined to form a complete trajectory Ω corresponding to the desired position point matrix P;

[0121] S4, optimize the initial trajectory Ω according to the analytical priority strategy;

[0122] S41, evaluating and updating the lower bound of the running time of each sub-trajectory according to the analytical method of the running time in the local asymmetric acceleration motion profile;

[0123] The running time of the i-th sub-trajectory of all joints is determined according to the following formula:

[0124]

[0125] Among them: et i represents the running time of the ith sub-trajectory, The running time of the i-th point-to-point trajectory type sub-trajectory of the g-th joint is represented by;

[0126] S42, optimizing the trajectory Ω using the obtained lower bound of the running time of each sub-trajectory, and determining whether the optimized trajectory Ω meets the target condition;

[0127] For working condition 1, the optimized trajectory can meet the target conditions; the trajectory simulation results generated by the optimal solution are as follows Figure 3 As shown, the six joint trajectories generated have continuous acceleration, without overshoot and undershoot;

[0128] For working condition 2, the optimized trajectory fails to meet the target conditions, so the numerical calculation phase is entered;

[0129] The objective function is,

[0130]

[0131] Among them, i,g is the overshoot of the i-th sub-trajectory of the g-th joint, u i,g is the undershoot of the i-th sub-trajectory of the g-th joint;

[0132] The constraints are,

[0133]

[0134] Among them, θ g is the position of the gth joint, is the speed limit, is the acceleration limit, is the jerk limit;

[0135] In said S42, the objective function includes trajectory running time, overshoot and undershoot, and the energy consumption and driving torque can be increased according to user needs; any multi-objective optimization algorithm and constraint control method are used to optimize the trajectory;

[0136] Target conditions include:

[0137] C1: at least one joint of the ith sub-trajectory is of point-to-point type;

[0138] C2: The generated complete trajectory does not have overshoot and undershoot;

[0139] C3: The generated complete trajectory satisfies the physical constraints of the robot.

[0140] S43, if the target conditions are met, the optimal solution is output and the trajectory planning is completed;

[0141] S44. If the target conditions are not met, the numerical calculation phase is entered until the termination conditions are met, and the optimal solution or Pareto optimal solution set is output to complete the trajectory planning.

[0142] The termination condition mainly refers to the number of iterations reaching the maximum iter_max; if the Pareto optimal solution set is obtained, the user selects the optimal solution according to the comprehensive fuzzy membership function of the Pareto frontier to generate the optimized trajectory. The process is as follows:

[0143] The multi-objective improved student psychology optimization algorithm and adaptive ε constraint technology are used for optimization, and the population is divided into four types of individuals: the best students, good students, ordinary students and students to be improved;

[0144] The best student is the best individual in the population, while the selection of good students, average students and students to be improved is random;

[0145] The best students' scores were improved using the following formula,

[0146]

[0147] in, is the best student individual at the iterth iteration, is the random student individual at the iter iteration, rand is a random number in [0,1], r 0 is a random number of 1 or 2;

[0148] The scores of good students are improved using the following two formulas, and the selection is also random:

[0149]

[0150] in, is the individual student whose score is the average, rand1 is a random number in [0,1], and rand2 is a random number in [0,1];

[0151] The score improvement of ordinary students adopts the following formula,

[0152]

[0153] The scores of students to be improved are improved using the following two formulas, and the selection is also random:

[0154]

[0155] Where, L(λ) is the Levy flight function;

[0156] The update of ε in the adaptive ε constraint technique is related to the number of feasible solutions and the current iteration number, as shown below:

[0157]

[0158] Among them, c f is the ratio of feasible solutions to all solutions in the current iteration, c iter It is the ratio of the current cumulative number of iterations to the maximum number of iterations, and iter_max is the maximum number of iterations;

[0159] The fast non-dominated sorting method is used to sort the individuals in the population, and the non-dominated individuals are selected and stored in the external memory;

[0160] When a new individual needs to be stored in the external memory, if the new individual is dominated by at least one individual in the external memory, then the new individual is not allowed to be placed in the external memory; if the new individual dominates existing individuals in the external memory, then these existing individuals are removed and the new individual is stored; if the new individual does not dominate existing individuals in the external memory and the existing individuals in the external memory do not dominate the new individual, then the new individual is stored in the external memory; if the new individual is allowed to be stored in the external memory and the number of individuals in the external memory has reached the upper limit, then individuals with low density are removed until the number of individuals in the external memory is equal to the upper limit;

[0161] The density of individuals is estimated using the following formula,

[0162]

[0163] Among them, D x (id) is the density of the idth individual in the decision space, D f (id) is the density of the idth element in the target space, and are the maximum and minimum values ​​of the j-th dimension variable, respectively. and are the maximum and minimum values ​​of the objective function of the jth dimension, respectively, and class_size is the population size;

[0164] Determine whether the termination condition is met. If not, continue to execute the multi-objective improved student psychology optimization algorithm; if it is met, terminate the calculation and output the Pareto optimal solution set;

[0165] The user selects the optimal solution based on the comprehensive fuzzy membership function of the Pareto front to generate the optimized trajectory;

[0166] s syn (id)=(s1(id)+s2(id)) / max{s1+s2};

[0167]

[0168] Among them, f j (id) is the jth objective function value corresponding to the idth optimal solution, s j (id) is the fuzzy membership function of the jth objective of the idth optimal solution, s syn (id) is the comprehensive membership value of the idth optimal solution.

[0169] Figure 4 It is the Pareto optimal solution set of a certain six-degree-of-freedom robotic arm of this application for working condition 2.

[0170] If the user selects the shortest running time as the highest priority, the optimized trajectory is as follows Figure 5 As shown, the generated six joint trajectories have slight overshoot and undershoot phenomena, but the running time is effectively reduced.

[0171] If the user selects the minimum sum of overshoot and undershoot as the highest priority, the optimized trajectory is as follows Figure 6 As shown, the generated six joint trajectories have no overshoot or undershoot although the running time is long.

[0172] The contents described in the embodiments of this specification are merely enumerations of implementation forms of the inventive concept and are for illustrative purposes only. The protection scope of the present invention should not be considered to be limited to the specific forms described in this embodiment, and the protection scope of the present invention also extends to equivalent technical means that can be thought of by ordinary technicians in this field based on the inventive concept.

Claims

1. A robot arm trajectory planning method based on structure-guided motion profile, characterized in that: The method comprises the following steps: S1, input the desired position point matrix P of the joint space and the physical constraints of the robot; S2, initializing parameters, the parameters including: slope coefficient and local asymmetric coefficient of the point-to-point trajectory model, variable boundary, population size, external memory size and maximum number of iterations; S3, generating an initial trajectory Ω based on the structure-guided motion profile; S4, optimize the initial trajectory Ω according to the analytical priority strategy; S5. Output the optimal solution or Pareto optimal solution set.

2. The robot arm trajectory planning method based on structure-guided motion profile according to claim 1, characterized in that: In S1, the expected position points of the input joint space are an n×m matrix, where n represents the number of joints and m represents the number of expected positions; the physical constraints of the robotic arm are set according to user needs, including joint angle constraints, joint velocity constraints, joint acceleration constraints and joint velocity and acceleration constraints.

3. The robot arm trajectory planning method based on structure-guided motion profile according to claim 1 or 2, characterized in that: In S3, the process of generating the initial trajectory Ω based on the structure-guided motion profile is: S31, generating a speed constraint attribute list corresponding to the expected position point matrix; S32, according to the generated speed constraint attribute list, splitting the expected position point matrix P into a plurality of multi-point expected position point matrices and a plurality of point-to-point expected position point matrices according to the joints at which they are located; S33, the multi-point expected position matrix and the point-to-point expected position matrix generate corresponding trajectories using the multi-point trajectory model and the point-to-point trajectory model respectively; S34. According to the position of the desired position point in the matrix, the obtained multi-point trajectory and the point-to-point trajectory are combined to form a complete trajectory Ω corresponding to the desired position point matrix P.

4. The robot arm trajectory planning method based on structure-guided motion profile according to claim 3, characterized in that: In said S31, the generated velocity constraint attribute list is a matrix of n×m, where n represents the number of joints and m represents the number of expected joint positions; Each value vc in the velocity constraint attribute list is determined by the movement direction of two adjacent sub-trajectories, and the calculation formula is: Among them, p i,g is the expected position of the i-th joint of the g-th joint, 1 represents the forward motion of the joint, -1 represents the reverse motion of the joint, and 0 represents the stop motion of the joint.

5. The robot arm trajectory planning method based on structure-guided motion profile according to claim 3, characterized in that: In the S32, the multi-point expected position point matrix is ​​a one-dimensional matrix, the number of joint expected position points exceeds 2, and the velocity constraint attribute values ​​of the start and end joint expected position points are both equal to 0; The point-to-point expected position matrix is ​​a one-dimensional matrix. The number of joint expected position points is equal to 2, and the velocity constraint attribute values ​​of the two joint expected position points are both equal to 0.

6. The robot arm trajectory planning method based on structure-guided motion profile according to claim 3, characterized in that: In said S33, any multi-point trajectory model and point-to-point trajectory model may be used to generate the corresponding trajectory, and after disassembly and reassembly, the complete trajectory model of each joint may be different; The point-to-point desired position matrix generates the corresponding trajectory using the local asymmetric acceleration motion profile, where the calculation formula for the local asymmetric acceleration motion profile is: Where J(t) is the time-varying acceleration, j peak is the peak jerk during joint motion, t i is the end time of the ith trajectory segment, h i is the running time of the i-th trajectory segment, t is the time variable; The displacement, velocity, and acceleration of the point-to-point trajectory are obtained by integrating the jerk; The multi-point expected position matrix uses the B-spline function to generate the corresponding trajectory, where the displacement θ is expressed by the quintic B-spline function as follows: Where: cp ni is the ni-th control point of the quintic B-spline function, u is the normalized knot vector, n cp is the total number of control points, N ni,5 (u) is the ni-th basis function; N ni,5 (u) is obtained using the de Boer-Cox recursion formula, Among them, u ni represents the nith node in the normalized node vector variable, N ni,j (u) is the ni-th j-th basis function, m1 and m2 are the two coefficients of the basis function; According to the position of the desired position point in the matrix, the obtained multi-point trajectory and point-to-point trajectory are combined to form a complete trajectory Ω corresponding to the desired position point matrix P; Structure-guided motion profiles can generate different complete trajectory models for different joints based on a list of velocity constraint attributes.

7. The robot arm trajectory planning method based on structure-guided motion profile according to claim 1 or 2, characterized in that: In S4, the point-to-point trajectory model is solved by an analytical method, and a unique and feasible solution can be obtained. The process of optimizing the initial trajectory Ω according to the analytical method priority strategy is: S41, evaluating and updating the lower bound of the running time of each sub-trajectory according to the point-to-point trajectory model used; S42, optimizing the trajectory Ω using the obtained lower bound of the running time of each sub-trajectory, and determining whether the optimized trajectory Ω meets the target condition; S43, if the target conditions are met, the optimal solution is output and the trajectory planning is completed; S44. If the target conditions are not met, the numerical calculation phase is entered until the termination conditions are met, and the optimal solution or Pareto optimal solution set is output to complete the trajectory planning.

8. The robot arm trajectory planning method based on structure-guided motion profile according to claim 7, characterized in that: In S41, the variable lower bounds set for the running time of all sub-trajectories are assigned, and the running time of the sub-trajectory of the point-to-point type is solved, and finally the running time of the i-th sub-trajectory of all joints is determined according to the following formula: Among them, et i represents the running time of the ith sub-trajectory, Represents the running time of the i-th point-to-point trajectory type sub-trajectory of the g-th joint.

9. The robot arm trajectory planning method based on structure-guided motion profile according to claim 7, characterized in that: In said S42, the objective function includes trajectory running time, overshoot and undershoot, and the energy consumption and driving torque can be increased according to user needs; any multi-objective optimization algorithm and constraint control method are used to optimize the trajectory; Target conditions include: C1: at least one joint of the ith sub-trajectory is of point-to-point type; C2: The generated complete trajectory does not have overshoot and undershoot; C3: The generated complete trajectory satisfies the physical constraints of the robot.

10. The robot arm trajectory planning method based on structure-guided motion profile according to claim 7, characterized in that: In S44, the termination condition mainly refers to that the number of iterations has reached the maximum iter_max; if the Pareto optimal solution set is obtained, the user selects the optimal solution according to the comprehensive fuzzy membership function of the Pareto frontier to generate the optimized trajectory, and the process is as follows: The multi-objective improved student psychology optimization algorithm and adaptive ε constraint technology are used for optimization. The population is divided into four categories of individuals: the best student, good student, ordinary student and students to be improved; the best student is the best individual in the population, and the selection of good students, ordinary students and students to be improved is random; The best students' scores were improved using the following formula, in, is the best student individual at the iterth iteration, is the random student individual at the iter iteration, rand is a random number in [0,1], r 0 is a random number of 1 or 2; The scores of good students are improved using the following two formulas, and the selection is also random: in, is the individual student whose score is the average, rand1 is a random number in [0,1], and rand2 is a random number in [0,1]; The score improvement of ordinary students adopts the following formula, The scores of students to be improved are improved using the following two formulas, and the selection is also random: Where, L(λ) is the Levy flight function; The update of ε in the adaptive ε constraint technique is related to the number of feasible solutions and the current iteration number, as shown below: Among them, c f is the ratio of feasible solutions to all solutions in the current iteration, c iter It is the ratio of the current cumulative number of iterations to the maximum number of iterations, and iter_max is the maximum number of iterations; The fast non-dominated sorting method is used to sort the individuals in the population, and the non-dominated individuals are selected and stored in the external memory; When a new individual needs to be stored in the external memory, if the new individual is dominated by at least one individual in the external memory, then the new individual is not allowed to be placed in the external memory; if the new individual dominates existing individuals in the external memory, then these existing individuals are removed and the new individual is stored; if the new individual does not dominate existing individuals in the external memory and the existing individuals in the external memory do not dominate the new individual, then the new individual is stored in the external memory; if the new individual is allowed to be stored in the external memory and the number of individuals in the external memory has reached the upper limit, then individuals with low density are removed until the number of individuals in the external memory is equal to the upper limit; The density of individuals is estimated using the following formula, Among them, D x (id) is the density of the idth individual in the decision space, D f (id) is the density of the idth element in the target space, and are the maximum and minimum values ​​of the j-th dimension variable, respectively. and are the maximum and minimum values ​​of the objective function of the jth dimension, respectively, and class_size is the population size; Determine whether the termination condition is met. If not, continue to execute the multi-objective improved student psychology optimization algorithm; if it is met, terminate the calculation and output the Pareto optimal solution set; The user selects the optimal solution based on the comprehensive fuzzy membership function of the Pareto front to generate the optimized trajectory; s syn (id)=(s1(id)+s2(id)) / max{s1+s2}; Among them, f j (id) is the jth objective function value corresponding to the idth optimal solution, s j (id) is the fuzzy membership function of the jth objective of the idth optimal solution, s syn (id) is the comprehensive membership value of the idth optimal solution.

Images