Multi-robot task allocation method based on variable neighborhood genetic algorithm
By combining the variable neighborhood genetic algorithm and the cube bounding box method, the problem of low task allocation efficiency in multi-robot systems is solved, achieving efficient and safe task allocation and path planning, and improving the quality and safety of task allocation in multi-robot systems.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- TIANJIN UNIVERSITY OF TECHNOLOGY
- Filing Date
- 2024-09-29
- Publication Date
- 2026-07-24
AI Technical Summary
In existing technologies for multi-robotic arm systems, traditional genetic algorithms struggle to obtain solutions closer to the optimal solution in complex work scenarios with multiple tasks and requirements. They are also prone to getting trapped in local optima, resulting in low task allocation efficiency and an inability to adapt to complex and ever-changing work environments.
A variable neighborhood genetic algorithm is adopted, which uses the sum of path costs as the objective function. Four neighborhood structures are designed to expand the search range. The cube bounding box method is combined to plan collision-free motion paths and dynamically select neighborhood structures for searching, thereby improving the global search capability and the local search capability.
It enables efficient, safe, and precise task allocation for multiple robotic arms in different working environments, improving the quality and safety of task allocation, reducing path costs, and enhancing convergence speed and allocation balance.
Smart Images
Figure CN119188746B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of robotic arm control technology, specifically relating to a multi-robotic arm task allocation method based on a variable neighborhood genetic algorithm. Background Technology
[0002] Currently, with the rapid maturation and development of automation technology, the application of multi-robotic arm systems in collaborative operations is becoming increasingly widespread. Facing complex work scenarios with multiple tasks and diverse requirements, multi-robotic arm systems offer advantages such as high work efficiency and low operating costs. However, in reality, under the influence of numerous factors such as variable tasks and uncertain environmental information, the solution set obtained using traditional genetic algorithms is mostly concentrated within a specific range. It is difficult to obtain a closer approximate solution globally based on the distribution of task points, and after a certain number of iterations, it will get stuck in a local optimum, unable to continue searching for better results. Therefore, how to rationally and efficiently allocate all tasks to different robotic arms has become one of the industry's technical challenges. Summary of the Invention
[0003] To address the shortcomings of existing technologies, this invention aims to propose a multi-robot task allocation method based on a variable neighborhood genetic algorithm. This method calculates the total path cost for all robotic arms to complete all (to be) executed tasks, using the total path cost as the objective function. The problem of solving the multi-robot task allocation problem is transformed into finding the sequence of execution tasks that minimizes the objective function value. This effectively solves the problems of low task allocation efficiency, poor allocation results, and inability to adapt to complex and changing working environments in existing multi-robot collaborative work technologies.
[0004] The objective of this invention is achieved through the following technical solution.
[0005] A multi-robot task allocation method based on a variable neighborhood genetic algorithm includes the following steps:
[0006] S1, make m robotic arms perform n tasks, the initial positions of the m robotic arms are the same, n>m, randomly set the order of the n tasks to obtain the task sequence, encode the task sequence based on the genetic algorithm to obtain the initial chromosome individuals, set the population size and perform population initialization on the initial chromosome individuals to obtain the initial chromosome population;
[0007] S2, iterate in the order of S2-1, S2-2, S2-3 and S2-4 until the number of iterations equals the maximum number of iterations, then terminate the iteration and output the optimal solution as the final optimal solution;
[0008] S2-1: Calculate the fitness function value of all chromosome individuals in the initial chromosome population, and perform a selection operation based on the fitness function value of the chromosome individuals to obtain the chromosome population after the selection operation; set the mutation probability and perform a mutation operation on the chromosome population after the selection operation based on the mutation probability to obtain the chromosome population after the mutation operation; calculate the fitness function value of all chromosome individuals in the chromosome population after the mutation operation, and take the chromosome individual with the largest fitness function value as the current optimal solution, where the fitness function value of each chromosome individual is the reciprocal of the objective function value of that chromosome individual;
[0009] S2-2, Determine the selected neighborhood structure: Determine the current iteration number. When it is the first iteration, set the initial weights of the four neighborhood structures to 1 and randomly select one of the four neighborhood structures as the selected neighborhood structure. When the iteration number is greater than 1, select the neighborhood structure with the largest weight and a weight less than 4 from the four neighborhood structures as the selected neighborhood structure.
[0010] S2-3: Based on the selected neighborhood structure, search for the current optimal solution and obtain the searched solution; treat the searched solution and the current optimal solution as chromosome individuals, calculate the objective function values of the searched solution and the current optimal solution respectively, and determine the relationship between the objective function values of the searched solution and the current optimal solution: if the objective function value of the searched solution is greater than or equal to the objective function value of the current optimal solution, discard the searched solution, output the current optimal solution as the optimal solution, and determine whether the weight of the selected neighborhood structure in this search is 1. If it is not 1, decrease the weight of the selected neighborhood structure in this search by 1; if the objective function value of the searched solution is less than the objective function value of the current optimal solution, take the searched solution as the optimal solution and output it, and determine whether the weight of the selected neighborhood structure in this search is 4. If it is not 4, increase the weight of the selected neighborhood structure in this search by 1.
[0011] S2-4, Determine if it is the maximum number of iterations: If it is the maximum number of iterations, terminate the iteration and output the optimal solution of this iteration as the final optimal solution; if it is not the maximum number of iterations, replace the chromosome individual with the smallest fitness function value in the initial chromosome population with the optimal solution and continue the iteration.
[0012] S3, randomly generate m-1 breakpoint values to divide the final optimal solution into m segments. Decode each segment of the final optimal solution from left to right, and obtain the set after decoding. For the i-th robotic arm R i The assigned sequence of tasks to be executed; set Q *The m task sequences are assigned to the corresponding robotic arms; each robotic arm performs the task in a clockwise direction and uses the cube bounding box method to complete collision-free motion path planning during the task execution.
[0013] In S2, the method for calculating the objective function value of a chromosome individual is as follows:
[0014] Randomly generate m-1 breakpoint values to divide the chromosome individual into m segments. Decode each segment of the chromosome individual from left to right. The decoded result is a set Q of m task sequences to be executed by the robotic arms, where Q = {Q...} i |i=1,…m}, where Q i For the i-th robotic arm R i The sequence of tasks to be executed, Q i Including g i One task, 1 <g i <n;
[0015] Calculate the R of the i-th robotic arm i Execution completes task sequence Q i Path cost C i C i The calculation formula is as follows:
[0016]
[0017] In the formula, R represents the i-th robotic arm i Execute task q i1 The robotic arm R i The initial position o moves to the task q i1 The path cost incurred during the process of finding the location. L i1 For the i-th robotic arm R i From initial position o to task q i1 The Euclidean distance to the location; Indicates robotic arm R i Execute task sequence Q i The path cost consumed during the (a+1)th task in the process. L i(a+1) For robotic arm R i From task q ia Location to Task q i(a+1) The Euclidean distance to the location; R represents the i-th robotic arm i implement Time by task The path cost incurred in moving from the current position to the initial position o. L oFor the i-th robotic arm R i From the task The Euclidean distance from the current position to the initial position o; where all of the above Euclidean distances can be derived from the task sequence Q. i Calculated.
[0018] Calculate the objective function value Cost. The formula for calculating Cost is:
[0019]
[0020] The formula for calculating the fitness function value (Fitness) is as follows:
[0021]
[0022] In S2, the selection operation adopts a tournament selection strategy, which involves repeating the following operation until the size of the chromosome population after the selection operation is the same as the size of the initial chromosome population: each time, a certain number of chromosome individuals are randomly selected from the initial chromosome population, and the optimal chromosome individual is selected to enter the chromosome population after the selection operation based on the fitness function value of the chromosome individual.
[0023] In S2, the mutation operation adopts the exchange mutation strategy. The exchange mutation strategy performs the mutation operation on the chromosome individuals determined to be mutated in the chromosome population after the selection operation as follows: randomly select two positions a and b in the chromosome individual, and then exchange the gene at position a with the gene at position b, while the genes at the other positions remain unchanged.
[0024] In S2, there are four neighborhood structures: two-point swap neighborhood structure, insertion neighborhood structure, reverse order neighborhood structure, and shuffled swap neighborhood structure.
[0025] In S1, a segmented sequence encoding method is used to encode the task sequence.
[0026] Compared with the prior art, the present invention has the following beneficial effects:
[0027] 1. This invention comprehensively considers the working state and task distribution of multiple robotic arms and proposes a multi-robot task allocation method based on a variable neighborhood genetic algorithm. This multi-robot task allocation method has the advantages of high efficiency, safety, reliability, and automation. The multi-robot task allocation method of this invention enables multiple robotic arms to complete tasks accurately and efficiently in different working environments, effectively solving the problems of low efficiency and complex and variable working environments in existing technologies for multi-robot task allocation. It is expected to be widely used in various fields such as industrial production, service, and scientific research.
[0028] 2. This invention comprehensively considers the motion paths of each robotic arm during the entire task execution process. In the task allocation process, the cube bounding box method is used to quickly calculate the collision-free motion paths of the multiple robotic arms, thereby enabling the multiple robotic arms to reach the task point smoothly and improving the safety and practicality of multiple robotic arms performing tasks.
[0029] 3. This invention improves upon the traditional genetic algorithm by introducing a variable neighborhood search algorithm. It designs four neighborhood structures to expand the search range of the variable neighborhood search algorithm and uses a weighted scoring mechanism to dynamically select different neighborhood structures for searching. This allows the initial chromosome population to evolve rapidly towards superior individuals, fully leveraging the global search capability of the genetic algorithm and the local search capability of the variable neighborhood search algorithm. This improves the convergence speed of multi-robot arms in task allocation and yields a higher quality final optimal solution.
[0030] 4. The present invention employs a variable neighborhood genetic algorithm to obtain better solution results. The path cost consumed when multiple robotic arms perform tasks is smaller and the task allocation is more balanced. At the same time, the variable neighborhood genetic algorithm has better convergence. The total objective function value (sum of path costs) of the multi-robotic arm multi-task allocation method based on the variable neighborhood genetic algorithm is reduced by 16.8% and 12.5% respectively compared with the genetic algorithm and the variable neighborhood search algorithm. This verifies that the multi-robotic arm multi-task allocation method based on the variable neighborhood genetic algorithm of the present invention is superior in solving the multi-robotic arm task allocation problem. Attached Figure Description
[0031] Figure 1 A comparison chart of iteration results for different algorithms;
[0032] Figure 2 A diagram showing the task allocation results for multiple robotic arms using the GA algorithm;
[0033] Figure 3 The result of multi-robot task allocation using the VNS algorithm is shown in the figure.
[0034] Figure 4 The diagram shows the task allocation results for multiple robotic arms using the GAVNS algorithm. Detailed Implementation
[0035] The following describes in detail, with reference to the accompanying drawings, a multi-robot task allocation method based on a variable neighborhood genetic algorithm according to the present invention.
[0036] In the following embodiments, the task allocation problem of multiple robotic arms in collaborative operations is: within the workspace of multiple robotic arms, to efficiently and reasonably allocate the execution tasks consisting of n spherical objects (small balls).
[0037] Assumptions: Each independent robotic arm is isomorphic and has the same execution capability; each spherical object can only be completed by one robotic arm at a time during the execution of the task.
[0038] Example 1
[0039] A multi-robot task allocation method based on a variable neighborhood genetic algorithm includes the following steps:
[0040] S1, make m robotic arms perform n tasks, the initial positions of the m robotic arms are the same, n>m, randomly set the order of the n tasks to obtain the task sequence, encode the task sequence based on the genetic algorithm to obtain the initial chromosome individuals, set the population size and perform population initialization on the initial chromosome individuals to obtain the initial chromosome population;
[0041] In S1, a segmented sequence encoding method (Wen Xinrong. Research on Task Allocation and Path Planning of Multi-Mobile Robots Based on Intelligent Warehouse [D]. Donghua University, 2023.) is used to encode the task sequence. The segmented sequence encoding method is not only simple to operate, but also effectively handles optimization problems with sequential relationships, ensuring that the work of each robotic arm is more balanced and reasonable, and improving the quality of the initial chromosome individuals. In this embodiment, the population size is set to 100.
[0042] S2, iterates in the order of S2-1, S2-2, S2-3 and S2-4 until the number of iterations equals the maximum number of iterations, then terminates the iteration and outputs the optimal solution as the final optimal solution; in this embodiment, the maximum number of iterations is set to 500.
[0043] S2-1: Calculate the fitness function value of all chromosome individuals in the initial chromosome population, and perform a selection operation based on the fitness function value of the chromosome individuals to obtain the chromosome population after the selection operation; set the mutation probability and perform a mutation operation on the chromosome population after the selection operation based on the mutation probability to obtain the chromosome population after the mutation operation; calculate the fitness function value of all chromosome individuals in the chromosome population after the mutation operation, and take the chromosome individual with the largest fitness function value as the current optimal solution, where the fitness function value of each chromosome individual is the reciprocal of the objective function value of that chromosome individual;
[0044] S2-2, Determine the selected neighborhood structure: Determine the current iteration number. When it is the first iteration, set the initial weights of the four neighborhood structures to 1 and randomly select one of the four neighborhood structures as the selected neighborhood structure. When the iteration number is greater than 1, select the neighborhood structure with the largest weight and a weight less than 4 from the four neighborhood structures as the selected neighborhood structure.
[0045] S2-3: Based on the selected neighborhood structure, search for the current optimal solution and obtain the searched solution; treat the searched solution and the current optimal solution as chromosome individuals, calculate the objective function values of the searched solution and the current optimal solution respectively, and determine the relationship between the objective function values of the searched solution and the current optimal solution: if the objective function value of the searched solution is greater than or equal to the objective function value of the current optimal solution, discard the searched solution, output the current optimal solution as the optimal solution, and determine whether the weight of the selected neighborhood structure in this search is 1. If it is not 1, decrement the weight of the selected neighborhood structure in this search by 1 (if it is 1, do not change the weight of the selected neighborhood structure in this search). The search selects the neighborhood structure weights (does not perform any operations); if the objective function value of the searched solution is less than the objective function value of the current optimal solution, then the searched solution is taken as the optimal solution and output, and it is determined whether the weight of the selected neighborhood structure in this search is 4. If it is not 4, the weight of the selected neighborhood structure in this search is incremented by 1 (if it is 4, no operation is performed on the weight of the selected neighborhood structure in this search); that is, steps S2-3 determine the selected neighborhood structure and search according to the weighted scoring mechanism, making the variable neighborhood search algorithm more efficient and flexible, and effectively improving the local search ability, flexibility and efficiency of the variable neighborhood search algorithm;
[0046] S2-4, Determine if it is the maximum number of iterations: If it is the maximum number of iterations, terminate the iteration and output the optimal solution of this iteration as the final optimal solution; if it is not the maximum number of iterations, replace the chromosome individual with the smallest fitness function value in the initial chromosome population with the optimal solution and continue the iteration.
[0047] In S2, the method for calculating the objective function value of a chromosome individual is as follows:
[0048] Randomly generate m-1 breakpoint values to divide the chromosome individual into m segments. Decode each segment of the chromosome individual from left to right. The decoded result is a set Q of m task sequences to be executed by the robotic arms, where Q = {Q...} i |i=1,…m}, where Q i For the i-th robotic arm R i The sequence of tasks to be executed, Q i Including g i One task, 1 <g i <n;
[0049] Calculate the R of the i-th robotic arm i Execution completes task sequence Q i Path cost C i C i The calculation formula is as follows:
[0050]
[0051] In the formula, R represents the i-th robotic arm i Execute task q i1 The robotic arm R i The initial position o moves to the task q i1 The path cost incurred during the process of finding the location. L i1 For the i-th robotic arm R i From initial position o to task q i1 The Euclidean distance to the location; Indicates robotic arm R i Execute task sequence Q i The path cost consumed during the (a+1)th task (robotic arm R) i By q ia The task location moves to q i(a+1) The path cost incurred at the location of the task. L i(a+1) For robotic arm R i From task q ia Location to Task q i(a+1) The Euclidean distance to the location; R represents the i-th robotic arm i implement Time by task The path cost incurred in moving from the current position to the initial position o. L o For the i-th robotic arm R i From the task The Euclidean distance from the current position to the initial position o.
[0052] Calculate the objective function value Cost (the sum of path costs for m robotic arms performing n tasks). The formula for calculating Cost is:
[0053]
[0054] The formula for calculating the fitness function value (Fitness) is as follows:
[0055]
[0056] In S2, the selection operation adopts a tournament selection strategy. The tournament selection strategy is to repeat the following operation until the size of the chromosome population after the selection operation is the same as the size of the initial chromosome population: each time, a certain number of chromosome individuals are randomly selected from the initial chromosome population (in this embodiment, the certain number is set to 10), and the optimal chromosome individual is selected to enter the chromosome population after the selection operation based on the fitness function value of the chromosome individual.
[0057] In S2, the mutation operation employs a crossover mutation strategy. This strategy involves randomly selecting two positions, a and b, within the chromosome population after the initial mutation. The gene at position a is then exchanged with the gene at position b, while the genes at other positions remain unchanged. This mutation operation improves the diversity of the chromosome population after the mutation and enhances the local search capability of the genetic algorithm. In this embodiment, the mutation probability is set to 0.05.
[0058] In S2, there are four neighborhood structures: two-point swap neighborhood structure, insertion neighborhood structure, reverse order neighborhood structure, and shuffled swap neighborhood structure.
[0059] The variable neighborhood search algorithm expands the search range by using four neighborhood structures, and by increasing the diversity of neighborhood structures, the objective function (total path cost) can find the final optimal solution.
[0060] S3, randomly generate m-1 breakpoint values to divide the final optimal solution into m segments. Decode each segment of the final optimal solution from left to right, and obtain the set Q after decoding. * , For the i-th robotic arm R i The assigned sequence of tasks to be executed; set Q * The m task sequences are assigned to the corresponding robotic arms (the task sequences are assigned to the corresponding robotic arms). Assigned to robotic arm R i Each robotic arm performs tasks in a clockwise direction, and during task execution, it uses the cube bounding box method (Yue Xu, Zhou Haibo, Shao Yanpeng, et al. Improved artificial potential field method for obstacle avoidance planning of robotic arm joint space [J]. Mechanical Transmission, 2023, 47(10): 23-30.) to complete collision-free motion path planning. The cube bounding box method has a small computational load and short computation time, which meets the requirements of real-time control and improves the operating efficiency of the robotic arm.
[0061] In the multi-robot task allocation method of Example 1, the number of robotic arms m = 3 and the number of tasks n = 30.
[0062] Example 2
[0063] Based on the m robotic arms in Example 1 performing n tasks, the multi-robotic arm task allocation is performed according to the genetic algorithm (GA).
[0064] Example 3
[0065] Based on the m robotic arms performing n tasks in Example 1, the multi-robotic arm task allocation is performed according to the Variable Neighborhood Search (VNS) algorithm.
[0066] Under the same parameter conditions, the convergence speed and optimization ability of the Variable Neighborhood Genetic Algorithm (GAVNS), Genetic Algorithm (GA), and Variable Neighborhood Search Algorithm (VNS) are compared through simulation. The 30 tasks are represented by numbers 1 to 30. The iterative curves of the multi-robotic arm task allocation methods in Examples 1-3 are shown below. Figure 1 As shown, from Figure 1 It can be seen that the multi-robot task allocation method based on the Variable Neighborhood Genetic Algorithm (GAVNS) in Embodiment 1 of the present invention has a faster convergence speed, and the objective function value is the lowest when the number of iterations reaches the maximum, indicating a stronger optimization ability.
[0067] The simulation results of the multi-robotic arm task allocation methods in Examples 1-3 are as follows: Figure 2 , Figure 3 , Figure 4 As shown in Table 1, the comparison results of the three multi-robot task allocation methods are presented.
[0068] Table 1. Multi-robotic arm task execution sequence table
[0069]
[0070]
[0071] As can be seen from the above data, the objective function value (total path cost) of the task sequence solved by the multi-robot task allocation method based on the variable neighborhood genetic algorithm of the present invention is reduced by 16.8% and 12.5% respectively compared with the genetic algorithm and the variable neighborhood search algorithm. The multi-robot task allocation method based on the variable neighborhood genetic algorithm (GAVNS) has the smallest objective function value, which is 5384.8.
[0072] The present invention has been described above by way of example. It should be noted that any simple modifications, alterations or other equivalent substitutions that can be made by those skilled in the art without creative effort without departing from the core of the present invention fall within the protection scope of the present invention.
Claims
1. A multi-robot task allocation method based on a variable neighborhood genetic algorithm, characterized in that, Includes the following steps: S1, make m robotic arms perform n tasks, the initial positions of the m robotic arms are the same, n>m, randomly set the order of the n tasks to obtain the task sequence, encode the task sequence based on the genetic algorithm to obtain the initial chromosome individuals, set the population size and perform population initialization on the initial chromosome individuals to obtain the initial chromosome population; S2, iterate in the order of S2-1, S2-2, S2-3 and S2-4 until the number of iterations equals the maximum number of iterations, then terminate the iteration and output the optimal solution as the final optimal solution; S2-1: Calculate the fitness function value of all chromosome individuals in the initial chromosome population, and perform a selection operation based on the fitness function value of the chromosome individuals to obtain the chromosome population after the selection operation; set the mutation probability and perform a mutation operation on the chromosome population after the selection operation based on the mutation probability to obtain the chromosome population after the mutation operation; calculate the fitness function value of all chromosome individuals in the chromosome population after the mutation operation, and take the chromosome individual with the largest fitness function value as the current optimal solution, where the fitness function value of each chromosome individual is the reciprocal of the objective function value of that chromosome individual; S2-2, Determine the selected neighborhood structure: Determine the current iteration number. When it is the first iteration, set the initial weights of the four neighborhood structures to 1, and randomly select one of the four neighborhood structures as the selected neighborhood structure. When the iteration number is greater than 1, select the neighborhood structure with the largest weight and a weight less than 4 from the four neighborhood structures as the selected neighborhood structure. S2-3: Based on the selected neighborhood structure, search for the current optimal solution and obtain the searched solution; treat the searched solution and the current optimal solution as chromosome individuals, calculate the objective function values of the searched solution and the current optimal solution respectively, and determine the relationship between the objective function values of the searched solution and the current optimal solution: if the objective function value of the searched solution is greater than or equal to the objective function value of the current optimal solution, discard the searched solution, output the current optimal solution as the optimal solution, and determine whether the weight of the selected neighborhood structure in this search is 1. If it is not 1, decrease the weight of the selected neighborhood structure in this search by 1; if the objective function value of the searched solution is less than the objective function value of the current optimal solution, take the searched solution as the optimal solution and output it, and determine whether the weight of the selected neighborhood structure in this search is 4. If it is not 4, increase the weight of the selected neighborhood structure in this search by 1. S2-4, Determine if it is the maximum number of iterations: If it is the maximum number of iterations, terminate the iteration and output the optimal solution of this iteration as the final optimal solution; if it is not the maximum number of iterations, replace the chromosome individual with the smallest fitness function value in the initial chromosome population with the optimal solution and continue the iteration. S3, randomly generate m-1 breakpoint values to divide the final optimal solution into m segments. Decode each segment of the final optimal solution from left to right, and obtain the set Q after decoding. * , For the i-th robotic arm R i The assigned sequence of tasks to be executed; set Q * The m task sequences are assigned to the corresponding robotic arms; each robotic arm performs the task in a clockwise direction and uses the cube bounding box method to complete collision-free motion path planning during the task execution.
2. The multi-robotic arm task allocation method according to claim 1, characterized in that, In S2, the method for calculating the objective function value of a chromosome individual is as follows: Randomly generate m-1 breakpoint values to divide the chromosome individual into m segments. Decode each segment of the chromosome individual from left to right. The decoded result is a set Q of m task sequences to be executed by the robotic arms, where Q = {Q...} i |i=1,…m}, where Q i For the i-th robotic arm R i The sequence of tasks to be executed, Q i Including g i One task, 1 <g i <n; Calculate the R of the i-th robotic arm i Execution completes task sequence Q i Path cost C i C i The calculation formula is as follows: In the formula, R represents the i-th robotic arm i From initial position o to task q i1 The Euclidean distance to the location; Indicates robotic arm R i From task q ia Location to task q i(a+1) The Euclidean distance to the location; R represents the i-th robotic arm i From the task The Euclidean distance from the current position to the initial position o; where all of the above Euclidean distances can be derived from the task sequence Q. i get; Calculate the objective function value Cost. The formula for calculating Cost is:
3. The multi-robotic arm task allocation method according to claim 1 or 2, characterized in that, The formula for calculating the fitness function value (Fitness) is as follows:
4. The multi-robotic arm task allocation method according to claim 3, characterized in that, In S2, the selection operation adopts a tournament selection strategy, which involves repeating the following operation until the size of the chromosome population after the selection operation is the same as the size of the initial chromosome population: each time, a certain number of chromosome individuals are randomly selected from the initial chromosome population, and the optimal chromosome individual is selected to enter the chromosome population after the selection operation based on the fitness function value of the chromosome individual.
5. The multi-robotic arm task allocation method according to claim 1, characterized in that, In S2, the mutation operation adopts the exchange mutation strategy. The exchange mutation strategy performs the mutation operation on the chromosome individuals determined to be mutated in the chromosome population after the selection operation as follows: randomly select two positions a and b in the chromosome individual, and then exchange the gene at position a with the gene at position b, while the genes at the other positions remain unchanged.
6. The multi-robotic arm task allocation method according to claim 1, characterized in that, In S2, there are four neighborhood structures: two-point swap neighborhood structure, insertion neighborhood structure, reverse order neighborhood structure, and shuffled swap neighborhood structure.
7. The multi-robotic arm task allocation method according to claim 1, characterized in that, In S1, a segmented sequence encoding method is used to encode the task sequence.