Consistency path planning method for multi-robot collaborative operation

Abstract

The invention belongs to the technical field of multi-robot path planning, and particularly relates to a consistent path planning method for multi-robot collaborative operation. Firstly, an environment map is converted into a grid map, and a starting point and a terminal point of each robot are set; then, calculating a priority score of each robot, layering all the robots according to the priority scores to obtain inter-layer priorities, and determining intra-layer priorities of the robots according to the priority scores; finally, according to the starting point and the ending point of the robot, a plurality of optional paths are generated for the robot through the Yen's algorithm, the motion paths of the robot are distributed according to the priority in the layer, and primary distribution of the motion paths is achieved; and according to the inter-layer priority, re-allocating the primarily allocated motion paths of the robots, and finely adjusting the primarily allocated motion paths to ensure that the motion paths of all the robots do not conflict. According to the method, a strategy of combining hierarchical path allocation, intra-layer priority ranking and inter-layer conflict detection is adopted, path conflicts can be effectively reduced, meanwhile, a new path generated through re-planning is continuation of a previous path or a subset of the previous path, continuity and consistency of path planning are guaranteed, and path changes are reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of multi-robot path planning, and in particular relates to a consistent path planning method for multi-robot collaborative operation. Background Art

[0002] With the development of industrial automation, multi-robot systems are increasingly used in industrial production, logistics and transportation. Multi-robots can improve production efficiency and reduce dependence on manpower through mutual collaboration, and have gradually become an important part of modern industry. In the process of multi-robot collaborative operation, the continuity and smooth transition of the path are very important. Sudden changes in the path may not only cause discontinuous robot movement, but also cause system instability, and even in some cases, may cause safety issues.

[0003] Existing path planning methods, such as the A* algorithm and the Dijkstra algorithm, can plan effective paths for a single robot, but in multi-robot scenarios, how to dynamically adjust the path to avoid collisions between robots is still a challenge. For multi-robot path planning, whether it is the WHCA* (Windowed Hierarchical Cooperative A*) algorithm or the CBS (Conflict-Based Search) algorithm, when a robot needs to adjust its path, it may affect the paths of other robots and produce a chain reaction, resulting in a decrease in system stability. When the path is interrupted by obstacles or other unexpected situations, it is usually necessary to re-plan a new path for the robot. However, the re-planning of the path often does not take into account the previously planned path information, resulting in a lack of continuity between the new path and the previous path, that is, the new path is not a natural continuation or subset of the previous path. The robot may have some behaviors that reduce travel efficiency due to path changes, such as turning and slowing down, thereby affecting the overall path planning efficiency and response speed. Therefore, how to effectively re-plan the path so that the newly generated path can make the most of the existing path information is a key challenge in path planning. Summary of the invention

[0004] In view of the deficiencies in the prior art, the technical problem that the present invention intends to solve is to provide a consistent path planning method for multi-robot collaborative operations.

[0005] The present invention solves the technical problem by adopting the following technical solution: A consistent path planning method for multi-robot collaborative operation, characterized in that the method comprises the following steps: Step 1: Convert the environment map into a grid map and set the starting and ending points of each robot on the grid map; Step 2: Set the robot's priority and calculate the priority score according to the following formula:

[0006] In the formula, is the priority score; is the task importance coefficient, Is the battery percentage, is the current priority, , and are all weight factors; All robots are layered according to the priority scores to obtain inter-layer priorities, each of which corresponds to a priority score interval; at the same time, the robot's intra-layer priority is determined according to the priority scores; Step 3: Based on the starting point and end point of the robot, use Yen's algorithm to generate multiple optional paths for the robot and assign the robot's movement path according to priority; First, for all robots corresponding to the inter-layer priority, motion paths are allocated to each robot in turn according to the intra-layer priority, that is, the robot with the highest priority in the layer selects the shortest optional path as the motion path; the robot with the second highest priority in the layer performs conflict detection on multiple optional paths according to the alternative sorting and the motion path of the robot with the highest priority in the layer, and selects the optional path without conflict as the motion path of the robot with the second highest priority in the layer; according to the above operations, motion paths are allocated to the remaining robots in turn to complete the initial allocation of motion paths; then, the motion paths initially allocated to the robots are allocated again according to the inter-layer priority, that is, the motion paths initially allocated to each robot with the second highest inter-layer priority are detected in turn with the motion paths of each robot with the highest inter-layer priority, and if there is a conflict, the remaining optional paths are selected until a non-conflicting motion path is allocated; The conflict detection process is as follows: perform cubic spline interpolation on the robot's path to obtain the interpolation equation of the path; make the interpolation equations of the two paths equal, and determine whether the two paths have an intersection by solving the equation group. If there is no intersection, it indicates that there will be no conflict, so the two paths do not conflict; if there is an intersection, it indicates that there may be a conflict in space, and then determine whether there is a conflict in time according to the following formula. If the following formula is true, it is considered that the two paths will conflict in time, so the two paths conflict; if the following formula is not true, it is considered that the two paths do not conflict in time, so the two paths do not conflict;

[0007] In the formula, , is the integrated distance from the starting point to the intersection of the two robots, , are the movement rates of the two robots, is the time conflict threshold.

[0008] Compared with the prior art, the present invention has the following beneficial effects: 1. Different from the conventional image conversion algorithm (such as OpenCV) to construct a raster map, the present invention adopts the MISSFormer model to perform semantic segmentation on the environment map to obtain the category label of each pixel; the category labels of all pixels in the grid unit are aggregated to obtain the category label of the grid unit, thereby realizing the pixel-level conversion of the map. Especially in complex or dynamic environments, it can not only accurately identify obstacles and free spaces, but also provide rich semantic information for other objects (such as vehicles, pedestrians, etc.), which can significantly improve the accuracy of converting the environment map into a grid map, and is conducive to improving the accuracy of path planning and conflict detection.

[0009] 2. Determine the priority of the robot based on the importance of the task, power and current priority, including inter-layer priority and intra-layer priority. Allocate motion paths for all robots in the layer for the first time according to the intra-layer priority, and then reallocate the motion paths of the robots according to the inter-layer priority. The strategy of combining hierarchical path allocation, intra-layer priority sorting and inter-layer conflict detection can effectively reduce path conflicts. When multiple robots suddenly encounter obstacles and stop while moving along the planned path, and re-plan the path again, most of the newly planned paths are part of the planned path before the last stop. The newly generated path is not only a continuation of the previous path, but also a subset of it as much as possible, thereby ensuring the continuity and consistency of the path, which not only reduces the additional computational burden caused by path changes, but also improves the robot's adaptability and response speed to environmental changes.

[0010] 3. The discrete path points are converted into continuous path functions using cubic spline interpolation to detect whether there is a conflict in space between the two optional paths of the robots. At the same time, by comparing the time difference with the time conflict threshold, it is further determined whether the two optional paths conflict in time, thereby achieving accurate detection of the conflict.

[0011] 4. Based on the shortest path, Yen's algorithm further generates multiple optional paths for each robot. When the robot's preferred path conflicts, it can switch to other optional paths to reduce path changes.

[0012] 5. The robot's priority changes dynamically and can be adjusted dynamically according to actual conditions, especially in complex scenarios, to maximize the overall system's operating efficiency while avoiding conflicts. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1is an overall flow chart of the method of the present invention; Figure 2 is the initial path planning result of the method of the present invention; Figure 3 It is the initial path planning result of CBS algorithm; Figure 4 is the secondary path planning result of the method of the present invention; Figure 5 It is the secondary path planning result of CBS algorithm. DETAILED DESCRIPTION

[0014] Specific embodiments are given below in conjunction with the accompanying drawings. The specific embodiments are only used to introduce the technical solutions of the present invention in detail and are not intended to limit the protection scope of the present application.

[0015] The present invention provides a consistent path planning method for multi-robot collaborative operation (hereinafter referred to as method, see Figures 1 to 5 ), including the following steps: Step 1: Convert the environment map into a grid map and set the starting and ending points of each robot on the grid map; The environment map usually contains open areas for robot movement and some impassable obstacles, such as walls, machinery and equipment, or shelves for storing goods. The environment map can be a simplified representation of an actual industrial scene, such as the layout of a factory workshop, the floor plan of a logistics warehouse, etc.

[0016] The MISSFormer model is used to segment each pixel of the environment map, predict its category, and obtain the category label of each pixel; first, the probability of the pixel belonging to each category is calculated to obtain the probability distribution of the pixel; the pixel in the environment map I The probability distribution of is: ,in, Represents pixels Belongs to category The probability of represents the number of categories; then, the category corresponding to the maximum probability is selected as the category label, then the pixel The category labels are: ; The category labels of all pixels are output as a segmentation map, which has the same size as the environment map and contains one category label for each pixel.

[0017] The environment map is evenly divided into multiple grid cells, each grid cell corresponds to an area, and the grid cell contains all the pixels in the area; according to the segmentation map, the grid cells Aggregate the category labels of all pixels in to get the grid unit Category label ; Rasterize the environment map according to the category label of the network unit to obtain a grid map. For example, if the category label is an obstacle, the value of the network unit is 1; if the category label is free space, the value of the grid unit is 0. The grid map provides the basis for setting the starting and ending points of the robot, because each grid cell is a potential position or target position of the robot.

[0018] The robot's starting point represents the initial position of the robot at the beginning of the task. It is usually set manually by the user according to the actual task, and a traversable position in the grid map is selected. It should be noted that the starting point cannot be set on an obstacle, otherwise the robot will not be able to start moving. Ideally, the starting point should be set at a suitable location in the robot's work area, such as the loading and unloading point of a logistics warehouse, next to a machine in a factory workshop, or near a processing area. Especially in complex multi-robot scenarios, multiple robots may share the same or adjacent starting points. In this case, attention should be paid to possible collisions with other robots.

[0019] The robot's end point represents the target location that the robot needs to reach at the end of the task, and is also manually set by the user based on the actual task; for example, in a logistics warehouse, the end point may be the location of a cargo rack; in a factory workshop, the end point may be next to another processing machine or assembly line. The end point should not overlap with obstacles, and it is also necessary to avoid possible collisions with other robots. Especially in complex scenarios with multiple robots, multiple robots may reach the same end point at the same time, so it is necessary to reasonably schedule them through priority and path planning to avoid congestion or conflicts at the end point.

[0020] Step 2: Set the robot's priority, including inter-layer priority and intra-layer priority; In multi-robot task scenarios, paths often cross or conflict between robots. Therefore, a priority is set for each robot. Robots with high priorities will have more freedom in path selection and can give priority to the optimal path, while robots with low priorities need to make adjustments based on the existing path selection results. Priorities are set based on a variety of factors, such as the urgency, importance, execution time, and robot model of the task. In some application scenarios, users may manually set priorities based on the importance of the task, such as assigning high-importance tasks to high-priority robots to ensure that high-importance tasks can be completed first. The present invention takes into account the robot's task importance, power, and current priority, and calculates the priority score according to the following formula:

[0021] In the formula, is the priority score; is the task importance coefficient, ranging from 0 to 1, which is used to characterize the importance of the task. The importance of the task directly affects the priority of the robot. The more important the task, the higher the priority; It is the percentage of battery power. The lower the battery power, the higher the robot priority. The higher the battery power, the lower the robot priority. is the current priority, with a value range of 1 to N, where N is the number of robots. The lower the current priority, the greater the impact on the priority score; , and are all weight factors; All robots are layered according to the priority scores to obtain inter-layer priorities; each inter-layer priority corresponds to a priority score interval, so each inter-layer priority contains multiple robots; at the same time, the robot's intra-layer priority is determined according to the priority score, and the higher the priority score, the higher the intra-layer priority.

[0022] As can be seen from the above formula, the priority of the robot is not static, but is dynamically adjusted by real-time evaluation of the importance of the robot's task and the current working status (such as remaining power, path occupancy, etc.). For example, a robot with low power may be given a higher priority to ensure that its motion path is not delayed; a robot with a high priority can be lowered after completing the task to optimize the collaborative efficiency of the robots in the system. When a robot stops due to encountering an obstacle or other unforeseen reasons, before the next path planning, the priority of the robot can be lowered so that other robots are given priority in allocating motion paths to ensure that the path planning result is as close as possible to the last planning result, so that the motion paths of the remaining robots, except for the stopped robot, can maintain the greatest degree of consistency, avoid unnecessary path adjustments as much as possible, ensure the efficiency and stability of the system, and reduce the resource consumption and complexity caused by frequent path changes.

[0023] In order to ensure the flexibility and robustness of the multi-robot system, the priority of the robots can be adjusted dynamically in real time according to the actual situation; for example, if a robot with a higher priority stagnates or has a task delay, its priority can be temporarily lowered to allow other robots to pass through to improve the overall efficiency of the system. On the contrary, if a robot has an urgent task, the priority of the robot can be increased to ensure that it can complete the task in a short time.

[0024] Step 3: Use Yen's algorithm to generate multiple optional paths for each robot and assign motion paths to the robots based on priority; In multi-robot tasks, conflicts may arise between paths, especially when multiple robots pass through the same area at the same time. Therefore, a single path often cannot meet the needs. Based on the starting and ending points of the robots, Yen's algorithm is used to generate multiple optional paths for the robots. When the preferred optional path is occupied or a conflict occurs, it can be flexibly switched to other optional paths, which is also an important means to resolve conflicts and improve efficiency.

[0025] First, in each inter-layer priority, motion paths are allocated to each robot in order according to the intra-layer priority. For the robot with the highest priority in the layer, the shortest optional path is directly selected as the motion path, because no other robot in the layer occupies any path in the environment; for the robot with the second highest priority in the layer, multiple optional paths are sorted according to the alternatives to perform conflict detection with the motion path of the robot with the highest priority in the layer, and the optional path without conflict is selected as the motion path of the robot with the second highest priority in the layer; according to the above operation, motion paths are allocated to the remaining robots in turn, and the robots with low priorities in the layer need to perform conflict detection of motion paths with the robots with high priorities in the layer one by one to complete the initial allocation of motion paths; then, the motion paths initially allocated to the robots are allocated again according to the inter-layer priority, that is, the motion paths initially allocated are fine-tuned, that is, the motion paths initially allocated to each robot with the second highest priority in the layer are sequentially conflict detected with the motion paths of each robot with the highest priority in the layer, and if there is a conflict, the remaining optional paths are selected; the robots with low inter-layer priorities need to perform conflict detection of motion paths with the robots with high inter-layer priorities one by one until non-conflicting motion paths are allocated. Different from the conventional motion path allocation strategy based on priority, the present invention expands the existing serial priority path allocation strategy into a combination of hierarchical path allocation, intra-layer priority sorting and inter-layer conflict detection, which can effectively reduce path conflicts between different layers and improve computational efficiency.

[0026] Path conflict detection is one of the key steps to ensure the safe operation of robots. When multiple robots move in the same environment, it is crucial to avoid conflicts, especially when they share part of the path or the paths are close. In order to accurately detect whether the robot may conflict during movement, the robot's path is interpolated with cubic spline to obtain the interpolation equation of the path; the interpolation equations of the two paths are made equal, and the two paths are judged by solving the equation group to determine whether there is an intersection. If there is no intersection, it indicates that there will be no conflict in space; if there is an intersection, it indicates that there may be a conflict in space, and it is necessary to further judge whether there is a conflict in time, that is, calculate the integral distance from the starting point to the intersection of the two robots respectively. Due to the different movement rates of the robots, the same path may also produce a large time difference at the intersection. Therefore, the movement rate is introduced to calculate the time difference of the two robots. If the time difference is less than the time conflict threshold, it is considered that the two paths will conflict in time. If the time difference is greater than or equal to the time conflict threshold, it is considered that the two paths will not conflict in time. The condition for conflict is:

[0027] In the formula, , is the integrated distance from the starting point to the intersection of robots A and B, , is the movement speed of robots A and B, is the time conflict threshold.

[0028] In some complex scenarios, the optional paths generated by a robot with a lower priority may conflict with the paths of other robots. In this case, the path of the robot can be dynamically replanned, that is, the number of optional paths can be increased until an optional path without conflict appears. In extreme cases, the priority of the robot can be adjusted dynamically. For example, if a robot with a lower priority cannot find an optional path without conflict for a long time, its priority can be increased to give it priority in the allocation of movement paths.

[0029] In order to verify the superiority of the method of the present invention, the method of the present invention is compared with the CBS algorithm. The initial path planning results of the method of the present invention and the CBS algorithm are shown in Figure 2 and 3 , which contains the motion paths of five robots in total (one color line represents the motion path of one robot, and the darker end is the starting point). In both methods, the starting and ending positions of each robot are the same. The intersection of all robot motion paths at a certain intermediate moment in the initial path planning result is used as the starting point, and the end point remains unchanged for secondary path planning. The secondary path planning results of the method of the present invention and the CBS algorithm are shown in Figure 4 and 5. Experimental results show that the method of the present invention finds a reasonable motion path in the initial path planning, and the path generated in the secondary planning is a subset of the initial path, showing consistency, ensuring the stability and continuity of the path. Although the path planning results of the CBS algorithm are reliable, they are not consistent and have a large computational overhead. The consistency characteristics of the method of the present invention make it perform particularly well in complex scenarios where frequent adjustments to the robot path are required. It can provide more stable and efficient path planning for multi-robot systems, has high adaptability and flexibility, and can effectively reduce path conflicts and improve path allocation efficiency.

[0030] Any matters not described in the present invention are applicable to the prior art.

Claims

1. A consistent path planning method for multi-robot collaborative operation, characterized in that: The method comprises the following steps: Step 1: Convert the environment map into a grid map and set the starting and ending points of each robot on the grid map; Step 2: Set the robot's priority and calculate the priority score according to the following formula: In the formula, is the priority score; is the task importance coefficient, Is the battery percentage, is the current priority, , and are all weight factors; All robots are layered according to the priority scores to obtain inter-layer priorities, each of which corresponds to a priority score interval; at the same time, the robot's intra-layer priority is determined according to the priority scores; Step 3: Based on the starting point and end point of the robot, use Yen's algorithm to generate multiple optional paths for the robot and assign the robot's movement path according to priority; First, for all robots corresponding to the inter-layer priority, motion paths are allocated to each robot in turn according to the intra-layer priority, that is, the robot with the highest priority in the layer selects the shortest optional path as the motion path; the robot with the second highest priority in the layer performs conflict detection on multiple optional paths according to the alternative sorting and the motion path of the robot with the highest priority in the layer, and selects the optional path without conflict as the motion path of the robot with the second highest priority in the layer; according to the above operations, motion paths are allocated to the remaining robots in turn to complete the initial allocation of motion paths; then, the motion paths initially allocated to the robots are allocated again according to the inter-layer priority, that is, the motion paths initially allocated to each robot with the second highest inter-layer priority are detected in turn with the motion paths of each robot with the highest inter-layer priority, and if there is a conflict, the remaining optional paths are selected until a non-conflicting motion path is allocated; The conflict detection process is as follows: perform cubic spline interpolation on the robot's path to obtain the interpolation equation of the path; make the interpolation equations of the two paths equal, and determine whether the two paths have an intersection by solving the equation group. If there is no intersection, it indicates that there will be no conflict, so the two paths do not conflict; if there is an intersection, it indicates that there may be a conflict in space, and then determine whether there is a conflict in time according to the following formula. If the following formula is true, it is considered that the two paths will conflict in time, so the two paths conflict; if the following formula is not true, it is considered that the two paths do not conflict in time, so the two paths do not conflict; In the formula, , is the integrated distance from the starting point to the intersection of the two robots, , are the movement rates of the two robots, is the time conflict threshold.

2. The consistent path planning method for multi-robot collaborative operation according to claim 1 is characterized in that: The specific process of the first step is: use the MISSFormer model to segment each pixel of the environment map and obtain the category label of each pixel; first, calculate the probability that the pixel belongs to each category to obtain the probability distribution of the pixel; then, select the category corresponding to the maximum probability as the category label of the pixel; divide the environment map evenly into multiple grid units, aggregate the category labels of all pixels in each grid unit respectively, and obtain the category label of each grid unit; rasterize the environment map according to the category label of the network unit to obtain a raster map.

Images