Multi-mobile-robot path planning method based on swarm intelligence

Abstract

The invention relates to a multi-mobile-robot path planning method based on swarm intelligence, and belongs to the technical field of robot path planning. The method comprises the following steps: setting the size of a map, starting and target positions and colors of a robot, creating a preset grid map, and initializing an ant colony algorithm, a genetic algorithm and a pheromone system; iteratively searching paths for the mobile robots with different starting points at the same time through an ant colony algorithm; a genetic algorithm is used for optimizing the path, conflict detection and processing are carried out after the path is optimized, and it is ensured that the final path is free of conflicts; and when the maximum number of iterations is reached, outputting the shortest path that each robot arrives at the target node and no collision exists between the robots. According to the path planning method, unnecessary turning can be reduced, the convergence speed of path searching is obviously improved, robot conflicts can be avoided, and the actual requirements of multi-robot path planning are met.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of robot path planning, and relates to a multi-mobile robot path planning method based on a swarm intelligence algorithm. BACKGROUND

[0002] Mobile robots have attracted extensive attention and continuous investment from the academic and industrial communities due to their excellent environmental adaptability, flexible task execution, and broad application scenarios. Mobile robots are essentially complex mobile systems that integrate environmental perception, dynamic decision-making, motion control, path planning, group collaboration, and advanced technologies such as deep learning and swarm intelligence. They combine the cutting-edge achievements of mechanical engineering, electronic engineering, computer science, control science and engineering, communication engineering, and artificial intelligence, representing the development direction of high-end intelligent equipment. Currently, mobile robot technology has shown great application value and broad development prospects in many fields such as industrial production (e.g., automated logistics, flexible assembly), precision medicine (e.g., surgical assistance, material distribution), modern agriculture (e.g., automated planting, picking), intelligent warehousing, security inspection, and even special operations.

[0003] Thanks to continuous research and development efforts, the overall performance of single mobile robots, including perception accuracy, decision-making ability, motion control stability, and single-task execution efficiency, has been significantly improved, enabling them to handle increasingly complex environments and tasks. However, when faced with complex tasks that require large-scale, multi-link, high-time-efficiency, or collaborative completion in highly dynamic and complex environments (e.g., efficient picking at multiple target points in large warehouses, collaborative material transfer in complex factory environments, large-scale collaborative search and rescue after disasters, etc.), the capabilities of single robots are particularly limited. Limited by their physical load, energy endurance, computing resources, and perception range, single robots often struggle to efficiently complete such tasks independently within a limited time. This greatly restricts the in-depth application of mobile robot technology in scenarios requiring large-scale and high-efficiency operations.

[0004] To overcome the limitations of single robots and meet the growing demand for high efficiency and strong collaboration capabilities, multi-mobile robot systems have emerged. Multi-robot systems can significantly improve overall work efficiency, enhance system robustness, expand task coverage, and effectively solve complex problems that single robots cannot handle through task allocation, information sharing, and behavior coordination among multiple individuals. However, one of the core challenges in achieving efficient multi-robot collaboration is efficient and conflict-free path planning. This not only requires planning the optimal or suboptimal path for each robot from the starting point to the target point, but also must ensure that all robots can avoid collisions in real time during movement, and achieve optimal or efficient path coordination on a global level.

[0005] Swarm intelligence optimization algorithms, especially those inspired by the behavior of biological groups in nature, have shown unique potential in the field of multi-robot path planning due to their distributed, self-organizing, and parallel search characteristics. Among them, the Ant Colony Optimization (ACO) algorithm simulates the behavior of an ant colony searching for the optimal path through communication and cooperation by pheromone. With its outstanding global search ability, good environmental adaptability, and relatively simple implementation, it is widely used to solve such problems. When the classic ant colony algorithm is applied to multi-robot path planning, it can theoretically plan collision-free paths for multiple robots and has the ability to adapt to environmental changes to some extent.

[0006] However, multi-mobile robot path planning techniques based on the classic ant colony algorithm have shortcomings in terms of convergence speed, parameter robustness, path smoothness and optimality (close to Euclidean distance), efficient and intelligent conflict resolution ability, and effective fusion of global and local optimization ability, such as slow convergence speed, high parameter sensitivity, and jagged path. These defects seriously restrict the deployment efficiency and practicality of multi-mobile robot systems in actual complex task scenarios, so it is particularly important to improve the classic ant colony algorithm and integrate other algorithms to improve the efficiency of mobile robot path planning. SUMMARY

[0007] Therefore, the purpose of the present application is to provide a multi-mobile robot path planning method based on swarm intelligence, which improves the efficiency, accuracy, and robustness of path planning by improving the classic ant colony algorithm and using genetic algorithms to locally adjust the initial path planned by the ant colony algorithm.

[0008] To achieve the above purpose, the present application provides the following technical solutions:

[0009] A multi-mobile robot path planning method based on swarm intelligence, comprising:

[0010] Setting the map size, robot starting and target positions, and colors, creating a preset grid map, initializing the ant colony algorithm, genetic algorithm, and pheromone system;

[0011] Iteratively searching for paths for mobile robots with different starting points using the ant colony algorithm;

[0012] Optimizing the path using a genetic algorithm, and performing conflict detection and processing after optimizing the path to ensure that the final path is collision-free;

[0013] When the maximum number of iterations is reached, output the shortest path for each robot to reach the target node without collision between robots.

[0014] Further, the ant colony algorithm is used to simultaneously search paths for mobile robots with different starting points, and the method comprises the following steps:

[0015] The heuristic information and the comprehensive pheromone concentration of the neighborhood nodes of each ant are obtained;

[0016] The selection probability of the neighborhood nodes is calculated based on the heuristic information and the comprehensive pheromone concentration, and the next node is selected;

[0017] Invalid paths are filtered, path pairs are generated, and it is checked whether there is a conflict in the same position based on the grid coordinates and the time step, and the conflict is resolved through a priority allocation strategy and a pheromone punishment mechanism.

[0018] The comprehensive pheromone concentration is calculated by the following formula:

[0019]

[0020] GR=1+κ×a

[0021] P new =P base ×LR×GR

[0022] In the formula, LR represents the direction smoothness reward, κ represents the direction consistency reward coefficient, θ max represents the maximum angle that meets the reward condition, GR represents the global direction guidance reward, a represents the alignment degree of the moving direction and the target direction, P new represents the comprehensive pheromone concentration, and θ represents the included angle between the current moving direction and the previous moving direction, P base represents the basic pheromone concentration.

[0023] Further, the conflict resolution through the priority allocation strategy and the pheromone punishment mechanism comprises the following steps: determining the priority according to the robot ID, the high-priority robot does not adjust the path, and for the low-priority robot, a waiting step or a detour is inserted at the conflict time point; and the concentration is inhibited at the conflict position to guide the path to avoid the conflict area.

[0024] Further, the genetic algorithm is used to optimize the path, and the method comprises the following steps: calculating the path fitness of each robot, and the fitness is the reciprocal of the Euclidean distance from the path starting point to the end point;

[0025] The path with the highest fitness is taken as the elite path; the roulette wheel selection method is used to select the parent path from the path set, and the elite path is added to the end of the parent path list;

[0026] The path is subjected to cross and mutation operations according to the cross and mutation probabilities;

[0027] Detect whether the optimized path exists conflict, if exists conflict, adjust the path again according to the robot priority order, the robot with high priority keeps the original path, and the robot with low priority avoids the conflict area by inserting waiting step or detouring at the conflict time point;

[0028] The pheromone concentration is updated by pheromone evaporation rate, and the pheromone factor and heuristic factor are updated by cosine annealing algorithm.

[0029] Further, the cross operation is determined by cross probability; if not cross, the parent path is used as the child path; if cross operation is performed, the common point of the two parent paths is found, a common point is randomly selected as the cross point, two child paths are generated by exchanging the part after the cross point of the parent path, and the child path is verified whether it is effective, if effective, it is reserved, otherwise the parent path is used instead;

[0030] The child path generated by the cross operation is used as the parent path of the mutation operation, and the mutation operation is determined by mutation probability; if not mutation, the parent path generated by the cross operation is used as the child path; if mutation operation is performed, a new node is generated in the neighborhood of the previous node of the insertion position in the path, and the new path after mutation is verified whether it is effective, if effective, the original path is replaced, otherwise the original path is reserved. Wherein, the insertion position is between the second node and the penultimate node of the selected path.

[0031] Further, the pheromone factor and heuristic factor are updated by the following formula:

[0032]

[0033] In the formula, alpha and beta represent pheromone factor and heuristic information factor, alpha0 and beta0 represent initial value of pheromone weight and initial value of heuristic information weight respectively, h and H represent current iteration number and maximum iteration number respectively.

[0034] The application has the advantages that: the application designs a new fusion algorithm combining the global search ability of improved ant colony algorithm and the local adjustment ability of genetic algorithm for the robot path planning and obstacle avoidance problem between robots, improves the path diversity, enhances the adaptability of the algorithm to complex environment, and improves the path search efficiency.

[0035] The research method for multi-mobile robot path planning based on swarm intelligence can reduce unnecessary turning under the guidance of the direction consistency reward, obtain the optimal path equal to the Euclidean distance from the starting point to the end point, and obviously improve the convergence speed in the search process.

[0036] Additional advantages, objects, and features of the application will be set forth in part by the description that follows, and in part will become apparent to those skilled in the art upon examination of same, or can be learned by the practice of the application. The goals and other advantages of the application can be realized and attained by the structure particularly pointed out in the written description and claims hereof. BRIEF DESCRIPTION OF DRAWINGS

[0037] In order to make the objects, technical solutions and advantages of the present application clearer, the preferred embodiments of the present application will be described in detail below with reference to the drawings, in which:

[0038] Figure 1 The flow chart of the method for multi-mobile robot path planning based on swarm intelligence provided by the embodiment of the present application is shown in Figure 1.

[0039] Figure 2 The preset grid map is shown in Figure 2.

[0040] Figure 3 The optimal paths searched by the classical ant colony algorithm and the improved ant colony algorithm are shown in Figure 3.

[0041] Figure 4 The convergence curve change graphs of the classical ant colony algorithm and the improved ant colony algorithm are shown in Figure 4.

[0042] Figure 5 The path planning results of the multi-mobile robot system in the side collision situation are shown in Figure 5.

[0043] Figure 6 The path planning results of the multi-mobile robot system in the head-on collision situation are shown in Figure 6. DETAILED DESCRIPTION

[0044] The present application can be implemented or applied in other different specific embodiments, and various modifications or changes can be made to the details based on different views and applications without departing from the spirit of the present application. It should be noted that the drawings provided in the following embodiments only illustrate the basic concept of the present application in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0045] The drawings are only used for exemplary illustration, and the representation is only a schematic diagram, not a physical diagram, and cannot be understood as a limitation on the present application; in order to better illustrate the embodiments of the present application, some components in the drawings may be omitted, enlarged or reduced, and do not represent the actual product size; it is understandable for those skilled in the art that some known structures and their descriptions in the drawings may be omitted.

[0046] The same or similar reference numerals in the drawings of the embodiments of the present application correspond to the same or similar components; in the description of the present application, it should be understood that if the terms "upper", "lower", "left", "right", "front", "back" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, therefore the terms describing the positional relationship in the drawings are only used for exemplary illustration, and cannot be understood as a limitation on the present application, for those skilled in the art, the specific meaning of the above terms can be understood according to the specific circumstances.

[0047] An embodiment of the present application provides a multi-mobile robot path planning method based on swarm intelligence, as shown in Figure 1 The method comprises the following steps.

[0048] 1. Set the map size, robot starting and target positions, and colors, create a preset grid map, initialize the ant colony algorithm, genetic algorithm and pheromone system.

[0049] The grid method is used to construct the map, as shown in Figure 2 , wherein the black squares represent obstacles, the white squares represent feasible regions, and the coordinate mapping is as shown in Figure 2 .

[0050] The initialization of the ant colony algorithm and the genetic algorithm includes initializing the pheromone concentration of each grid of the grid map, and pre-setting the pheromone factor, heuristic factor, number of ants, genetic population size, crossover probability, mutation probability, decay coefficient, maximum iteration number, pheromone evaporation rate and direction consistency reward factor.

[0051] 2. Start the iterative search path, including:

[0052] 1) Initialize pheromone system, wherein the pheromone concentration of feasible region is 0.5, and the pheromone concentration of obstacle is 0.

[0053] 2) Simultaneously search paths for different starting ants until the path reaching the target node is found or the iteration exceeds the maximum number of steps. The process is as follows:

[0054] A1, obtain the heuristic information and the comprehensive pheromone concentration of the neighborhood nodes of each ant, wherein the comprehensive pheromone concentration is obtained according to the pheromone concentration of the neighborhood nodes, the angle size between the current moving direction and the last step moving direction, and the angle size between the target direction and the current direction, as shown in the following formula:

[0055]

[0056] GR = 1 + κ × a

[0057] P new = P base × LR × GR

[0058] In the formula, LR represents the direction smoothness reward; κ represents the direction consistency reward coefficient; θ max represents the maximum angle that meets the reward condition; GR represents the global direction guidance reward; a represents the alignment degree of the moving direction and the target direction, and the alignment degree is equal to the dot product of the target direction vector and the moving direction vector, and a reward is given when the alignment is positive; θ represents the angle between the current moving direction and the last step moving direction; P base represents the basic pheromone concentration. The above two reward methods act on the pheromone value in the form of product, and finally affect the comprehensive pheromone concentration of the ant transfer probability.

[0059] A2, calculate the selection probability of the neighborhood nodes based on the heuristic information and the comprehensive pheromone concentration, and select the next node, as shown in the following formula:

[0060]

[0061] In the formula, p i represents the selection probability of the i-th neighborhood node; P new,i represents the comprehensive pheromone concentration of the i-th neighborhood node; η i represents the heuristic information of the i-th neighborhood node; j represents the set of all neighborhood nodes, which is used to traverse all candidate nodes around the current position, and the probability is normalized by summation; α and β represent the pheromone factor and the heuristic information factor.

[0062] A3, filter invalid paths, generate path pairs, check for conflicts in the same location for each pair of paths at each time step, record conflict time and location and remove duplicates. Specifically, based on grid coordinates and time steps, determine whether the robot path is in conflict, and resolve conflicts through priority allocation strategy and pheromone punishment mechanism.

[0063] wherein the priority is determined according to the robot ID, and for robots with higher priority, their paths are not adjusted; for robots with lower priority, a waiting step or detour is inserted at the conflict time point. The insertion of the waiting step is to extend the moving trajectory of the low-priority robot by inserting repeated path nodes, forcing it to stop before the conflict position to stagger the time steps, achieving dynamic obstacle avoidance.

[0064] The pheromone punishment mechanism specifically applies a concentration inhibition to the conflict position, reducing the possibility of subsequent path selection in this area, guiding the path to avoid the conflict area.

[0065] A4, use genetic algorithm to optimize the path, improve the quality of the path through selection, crossover and mutation operations. After optimizing the path, conflict detection and processing are performed again to ensure that the final path is conflict-free.

[0066] wherein the candidate paths generated by the ant colony algorithm are locally optimized through selection, crossover and mutation operations of the genetic algorithm to improve path diversity and global search ability, and the optimization steps are as follows:

[0067] ① Calculate the path fitness of each robot, and the fitness is defined as the reciprocal of the Euclidean distance from the starting point to the end point, and the Euclidean distance is calculated as follows:

[0068]

[0069] wherein l represents the Euclidean distance from the starting point to the end point; t represents the tth step of the moving robot; T represents the total number of steps of the moving robot; X t represents the horizontal coordinate value of the moving robot at the tth step; X t-1 represents the horizontal coordinate value of the moving robot at the t-1th step; Y t represents the vertical coordinate value of the moving robot at the tth step; Y t-1 represents the vertical coordinate value of the moving robot at the t-1th step.

[0070] Find the path with the highest fitness as the elite path; use the roulette wheel selection method to select parent paths from the path set, and add the elite path to the end of the parent path list.

[0071] ②Decide whether to perform the crossover operation through the crossover probability. If not, directly use the parent path as the child path; if the crossover operation is performed, find the common points of the two parent paths, randomly select a common point as the crossover point, generate two child paths by exchanging the parts after the crossover point of the parent paths, and verify whether the child paths are valid. If valid, retain them; otherwise, use the parent path instead.

[0072] ③Decide whether to perform the mutation operation through the mutation probability. Use the child path generated by the crossover operation as the parent path of the mutation operation. If not, use the parent path generated by the crossover operation as the child path; if the mutation operation is performed, randomly select an insertion position in the path, which is required to be between the second node and the second-to-last node, generate a new node in the neighborhood of the node before the insertion position, and verify whether the new path after mutation is valid. If valid, replace the original path; otherwise, retain the original path.

[0073] ④Detect again whether there is a conflict in the optimized path. If there is a conflict, adjust the path again according to the priority order of the robots. For robots with higher priority, their paths are not adjusted; for robots with lower priority, insert a waiting step or detour at the conflict time point.

[0074] ⑤Update the pheromone concentration through the pheromone evaporation rate, and update the pheromone factor and heuristic factor through the cosine annealing algorithm.

[0075] wherein the pheromone factor and the heuristic factor are updated by the following formula:

[0076]

[0077] wherein a and β represent the pheromone factor and the heuristic factor, a0 and β0 represent the initial values of the pheromone weight and the initial value of the heuristic information weight respectively, h and H represent the current iteration number and the maximum iteration number respectively.

[0078] 3. When the iteration number reaches the maximum iteration number, output the shortest path of each robot reaching the target node without collision between the robots.

[0079] Based on the path planning method provided in the embodiment, MATLAB software is used to simulate it to verify the effectiveness and excellent effect of the path planning.

[0080] In the grid map shown in FIG. 2, the simulation verification is performed, and the parameters are set as follows: the pheromone factor is 1.5, the heuristic factor is 6, the number of ants is 10, the genetic population size is 10, the crossover probability is 0.4, the mutation probability is 0.1, the decay coefficient is 0.5, the maximum iteration number is 100, the pheromone evaporation rate is 0.2, and the direction consistency reward factor is 0.2.

[0081] The simulation results are as follows: Figure 3 , Figure 4 , Figure 5 , Figure 6 As shown.

[0082] like Figure 3 The diagram shows the optimal paths found by the classic ant colony algorithm and the improved ant colony algorithm. The red route on the left is the optimal path found by the improved ant colony algorithm in this embodiment, while the blue route on the right is the optimal path found by the classic ant colony algorithm. It can be seen that the path found by the improved ant colony algorithm in this embodiment reduces unnecessary turns and has a shorter path length than the classic ant colony algorithm.

[0083] Figure 4 The figure shows the convergence curves of the classic ant colony algorithm and the improved ant colony algorithm. Figure 4 The left side of the image shows the improved ant colony algorithm, while the right side shows the classic ant colony algorithm. It can be seen that the improved ant colony algorithm proposed in this embodiment significantly improves the convergence speed during the search process compared to the classic ant colony algorithm.

[0084] Figure 5 The path planning results of the multi-robot system under side collision conditions shown can be seen that the mobile robots avoid the intersection of their three-dimensional path state curves by avoiding conflict coordinate points, thus successfully realizing the obstacle avoidance function between robots.

[0085] Figure 6 The image shows the path planning results of a multi-robot system in a collision scenario. By stopping at the grid coordinates of the previous step, the robot successfully avoided colliding with other robots.

[0086] Simulation results show that the multi-robot path planning method based on swarm intelligence proposed in this invention can reduce unnecessary turns by relying on the guidance of directional consistency rewards, and obtain the optimal path equal to the Euclidean distance from the starting point to the destination. At the same time, the convergence speed is also significantly improved during the search process. Moreover, the improved ant colony algorithm that integrates genetic algorithm correctly handles different path conflict scenarios. Therefore, the fusion algorithm can effectively solve the problems of traditional algorithms easily converging to local optima and insufficient global path search capability. It can also effectively avoid robot conflicts in multi-robot systems and effectively meet the actual needs of multi-robot path planning.

[0087] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A multi-mobile robot path planning method based on swarm intelligence, characterized in that, The method includes: Set the map size, robot starting and target positions, and colors; create a preset grid map; and initialize the ant colony algorithm, genetic algorithm, and pheromone system. The ant colony algorithm iteratively searches for paths for mobile robots with different starting points simultaneously. Genetic algorithms are used to optimize the path, and after optimization, conflict detection and handling are performed to ensure that the final path is conflict-free. When the maximum number of iterations is reached, output the shortest path for each robot to reach the target node without collisions between robots.

2. The method according to claim 1, characterized in that, The ant colony algorithm iteratively searches for paths for mobile robots starting from different points simultaneously, including: Obtain heuristic information and comprehensive pheromone concentration of each ant's neighboring nodes; The selection probability of neighboring nodes is calculated based on heuristic information and comprehensive pheromone concentration, and the next node is selected. Invalid paths are filtered out, path pairs are generated, and for each path pair, the existence of conflicts with the same position is checked based on the raster coordinates and time step. Conflicts are resolved through a priority allocation strategy and a pheromone penalty mechanism.

3. The method according to claim 2, characterized in that, The overall pheromone concentration is calculated using the following formula: GR=1+κ×a P new =P base ×LR×GR In the formula, LR represents the directional smoothness reward; κ represents the directional consistency reward coefficient; θ max represents the maximum angle that satisfies the reward conditions; GR represents the global directional guidance reward; a represents the alignment between the movement direction and the target direction; P new P represents the overall pheromone concentration; θ represents the angle between the current movement direction and the previous movement direction. base This indicates the basic pheromone concentration.

4. The method according to claim 2, characterized in that, Conflict resolution through priority allocation strategies and pheromone penalty mechanisms includes: determining priority based on robot ID, with high-priority robots not adjusting their paths, and for low-priority robots, inserting waiting steps or detouring at the conflict point; applying concentration suppression at the conflict location to guide the path away from the conflict area.

5. The method according to claim 2, characterized in that, Optimizing paths using genetic algorithms includes: calculating the path fitness for each robot, where fitness is the reciprocal of the Euclidean distance from the start point to the end point of the path; The path with the highest fitness is designated as the elite path; a roulette wheel selection method is used to select parent paths from the path set, and the elite paths are added to the end of the parent path list. Perform crossover and mutation operations on the paths based on the crossover and mutation probabilities; If there is a conflict in the optimized path, the path is adjusted again according to the priority of the robots. Robots with higher priority keep their original paths, while robots with lower priority avoid the conflict area by inserting waiting steps or detouring at the conflict time point. The pheromone concentration is updated using the pheromone evaporation rate, and the pheromone factor and heuristic factor are updated using the cosine annealing algorithm.

6. The method according to claim 5, characterized in that, Whether to perform a crossover operation is determined by the crossover probability. If no crossover is performed, the parent path is used as the child path. If a crossover operation is performed, the common point of the two parent paths is found, and a common point is randomly selected as the crossover point. By swapping the part of the parent path after the crossover point, two child paths are generated, and the validity of the child paths is verified. If the child paths are valid, they are retained; otherwise, the parent paths are used instead. The child path generated by the crossover operation is used as the parent path of the mutation operation. The mutation probability determines whether to perform the mutation operation. If the mutation is not performed, the parent path generated by the crossover operation is used as the child path. If the mutation operation is performed, an insertion position is randomly selected in the path, and a new node is generated in the neighborhood of the node before the insertion position. The validity of the new path after mutation is verified. If it is valid, the original path is replaced; otherwise, the original path is retained.

7. The method according to claim 6, characterized in that, The insertion position is between the second node and the second-to-last node of the selected path.

8. The method according to claim 5, characterized in that, The pheromone factor and heuristic factor are updated using the following formula: In the formula, α and β represent the pheromone factor and the heuristic information factor, α0 and β0 represent the initial values ​​of the pheromone weight and the heuristic information weight, respectively, and h and H represent the current iteration number and the maximum iteration number, respectively.

Images