A flattening robot full coverage path planning method
Patent Information
- Application Number
- CN202510782553.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-12
- Publication Date
- 2026-10-30
- Estimated Expiration
- 2045-06-12
AI Technical Summary
但现有环境覆盖算法在复杂场景下存在的路径冗余、死点无法逃离及子区域覆盖遗漏等问题
[0038]1、本发明所提供的一种整平机器人全覆盖路径规划方法，局部区域覆盖算法优化：引入PMTS算法(Prioritize Moving Towards Starting)，定义八个移动方向(N、S、E、W、NE、NW、SE、SW)并设定三层优先级规则，引导机器人优先向起始点方向、水平方向运动，从而高效绕过障碍物并降低遗漏区域。
Smart Images
Figure CN120593760B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of robot path planning technology, specifically relating to a full-coverage path planning method for a leveling robot. Background Technology
[0002] In recent years, with the rapid development of artificial intelligence algorithms and application technologies, service mobile robots have been widely used in various scenarios. Leveling robots, as a typical example, have demonstrated significant advantages in ground construction and floor paving due to their automation and intelligence. Especially in indoor working environments, leveling robots can not only replace manual labor in completing heavy and precision-required leveling tasks, but also effectively improve construction efficiency and quality. However, due to the lack of stable Global Navigation Satellite System (GNSS) signals in indoor environments, the autonomous navigation and path planning of leveling robots still face many challenges. Therefore, researching efficient, stable, and adaptable path planning algorithms has become one of the key technologies driving the practical application and intelligent development of leveling robots.
[0003] Path planning for mobile robots can be mainly divided into two categories: The first is where the robot autonomously plans an optimal path from the starting point to the target point, requiring the shortest path, the least time, and ensuring safety and collision-free travel. The second is where the robot performs comprehensive traversal path planning in two-dimensional space, aiming to maximize operational efficiency, reduce path repetition, and achieve effective obstacle avoidance. Currently, numerous scholars both domestically and internationally have conducted in-depth research on mobile robot path planning, proposing and optimizing a large number of algorithms. These algorithms include traditional path planning methods such as A*, Dijkstra's algorithm, Rapid Expanding Random Tree (RRT), Probabilistic Roadmap (PRM), and Artificial Potential Field (APF), as well as intelligent optimization strategies such as Genetic Algorithm (GA), Artificial Fish Swarm Algorithm (AFS), Particle Swarm Optimization (PSO), and Ant Colony Optimization (ACO). Among the many path planning algorithms, the A* algorithm, as a classic heuristic search method, occupies an important position in mobile robot path planning research due to its high computational efficiency, strong path search capability, and guarantee of finding the optimal solution. The A* algorithm effectively guides the search process by combining actual and estimated costs, reducing unnecessary node expansion and improving search efficiency. Due to its clear algorithmic structure, simple implementation, and ease of integration with grid maps and graph search technologies, the A* algorithm has been widely applied in various path planning tasks, exhibiting excellent stability and reliability, especially in static environments, making it one of the most commonly used algorithms in robot autonomous navigation systems. However, existing environment coverage algorithms suffer from problems such as path redundancy, inability to escape dead points, and omissions in sub-region coverage in complex scenarios. Summary of the Invention
[0004] The purpose of this invention is to solve the above-mentioned problems and provide a full-coverage path planning method that combines multi-directional priority control strategy and dynamic A* path planning optimization, which is used to improve the operation efficiency and coverage integrity of mobile robots.
[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is: a method for full-coverage path planning of a leveling robot, comprising the following steps:
[0006] S1. Obtain an environmental raster map containing impassable areas;
[0007] S2. In the current sub-region, call the priority starting point movement local path planning algorithm, namely the PMTS algorithm, to guide the robot to complete the full coverage operation in the region until it reaches the dead point of the current sub-region.
[0008] S3. During the robot's movement, the robot checks the status of adjacent grid cells. If the coverage edge condition is met, the grid cells are added to the target point list.
[0009] S4. Eliminate unreasonable or redundant target points based on preset screening criteria to improve the efficiency of subsequent path generation.
[0010] S5. Select the target point with the closest Euclidean distance to the current dead point as the entry point of the next sub-region, and call the improved A* algorithm of the region connection path planning algorithm to generate the connection path;
[0011] S6. Proceed along the planned path to the target point, update the current location to that point, restart the partial coverage process, and repeat the process.
[0012] S7. If the current target point list is empty, it means that all reachable areas have been visited, and the coverage is complete.
[0013] Furthermore, the PMTS algorithm in S2 involves eight movement directions: east, south, west, north, southeast, northeast, southwest, and northwest. It also sets three priority rules for these directions. The first rule states that east, south, west, and north have higher priority than southeast, northeast, southwest, and northwest. The second rule indicates that if multiple feasible directions exist during path planning, the direction towards the starting position should be prioritized to effectively reduce path redundancy. The third rule further clarifies that in determining specific directions, horizontal directions have higher priority than vertical directions, which helps improve the overall efficiency and coherence of path planning.
[0014] Furthermore, during the local path planning process in S2, the robot scans the grid status in eight directions around it in real time, identifies uncovered grid cells that are passable, and adds them to the target point candidate list. Candidate target points must be located on the edge or boundary grid of an uncovered area and meet the predefined conditions of edge grids. If all grids surrounding a candidate point are covered areas or obstacles, it is considered to lack edge attributes and is removed to avoid redundant coverage and path breaks. Among the retained candidate target points, the target point with the smallest Euclidean distance to the current dead point is selected as the target position for the next connecting path, balancing path optimization and computational efficiency.
[0015] Furthermore, the regional connectivity path planning algorithm in S5 is a weighted average heuristic function method that combines Manhattan distance and Euclidean distance. Traditional A* algorithms expand some redundant nodes during path search, a phenomenon stemming from the inability of the heuristic function h^*(n) to flexibly adjust according to the real-time search process. When the estimated cost is lower than the actual cost, the number of nodes expanded by the algorithm increases significantly, leading to a decrease in search efficiency. In this case, the weight of h^*(n) needs to be increased to accelerate the calculation speed. When the estimated cost h^*(n) equals the actual cost, only necessary nodes are traversed. If the estimated cost is higher than the actual cost, although the search speed increases, the optimal path is easily missed; therefore, the weight should be reduced to balance path quality and efficiency.
[0016] Furthermore, the PMTS algorithm in S2 includes the following sub-steps:
[0017] S21. Initialize environment and parameters: Load the environmental space model M for the leveling robot, and initialize the current position current_pos, grid size l, step size ε, and local coverage path.
[0018] S22. Environmental Information Acquisition: Use sensors to acquire current environmental data, update the grid status and synchronously update the environmental spatial model M to ensure a true reflection of the environment and avoid path deviation or coverage omissions caused by lag in environmental information.
[0019] S23. Determine the movable direction: Determine whether movement is possible based on the grid status; free and uncovered grids are considered feasible directions, while covered or obstacle grids are considered infeasible directions; using the information obtained in step S22, determine all movable directions of the current position.
[0020] S24. Check feasible paths: Check if there is a direction that can be moved; if not, jump to step S27; if there is a feasible direction, proceed to the next step.
[0021] S25. Priority Direction Selection: According to the priority rules of the PMTS algorithm, the optimal direction is selected from all feasible directions, giving priority to the direction towards the starting point and the horizontal direction.
[0022] S26. Perform movement: The leveling robot moves once according to the selected direction and step size. After completion, it returns to step S22 to continue updating environmental information and making judgments.
[0023] S27. Complete Local Coverage: When the leveling robot reaches the end of the current area or a position where it can no longer move forward, the local coverage task ends and the path result Local Coverage Path is output.
[0024] Furthermore, the selection principle of the target point selection mechanism in S4 is as follows: a target point list is established. While performing local area coverage, the robot will scan the grid status in eight directions around it, select grid cells that have not yet been covered and are in a passable state, and record them in the target point list as candidate points for subsequent path planning. If a target point has been covered by the robot, it should be removed from the list immediately to ensure the accuracy of the data. At the same time, target points selected in the area connection path planning should also be removed from the list to optimize the distribution of target points.
[0025] Furthermore, in S4, removing unreasonable or redundant target points is the screening of target points. Target points should be selected from the edge positions of areas that are not yet covered. If the left, right, front, back, or diagonally adjacent grids of a node are simultaneously in a free and uncovered state, then the node can be considered as not having edge and corner characteristics.
[0026]
[0027] ρ(m)=μ(d1,d5)+μ(d2,c6)+μ(d3,d7)+μ(d4,d8)
[0028] When ρ(m)≥1, it means that the node is located in a continuous free region and does not have boundary features. Therefore, it is determined to be an invalid target point and should be removed from the candidate set. When ρ(m)=0, it means that there are no two grids that are simultaneously free and uncovered in its horizontal, vertical, or diagonal directions. In this case, the grid is located at the corner and can be retained as a reference target point for subsequent path planning.
[0029] Furthermore, the specific steps of the A* algorithm in S5 are as follows:
[0030] S51. Set the starting point as S and the ending point as G; initialize a list named setopenlist to store candidate nodes that have not yet been expanded, and create setclosedlist to record nodes that have been visited; the starting node S will be added to setopenlist first, while setclosedlist is still empty at this time, making preliminary preparations for the path search process;
[0031] S52. Expand the current parent node, add all its reachable adjacent child nodes to setopenlist, and move the parent node itself into setclosedlist, indicating that the node has been visited;
[0032] S53. Check if setopenlist is empty. If setopenlist is empty, it means that no feasible path could be found during the search, and the program terminates; otherwise, continue to the next step.
[0033] S54. Calculate the cost value of the current expandable child node according to the A* evaluation function, and select the child node with the smallest evaluation value as the new parent node.
[0034] S55. Determine whether the currently selected parent node is the target point G. If it is, return to the starting point from the target point by backtracking to generate the final path and end the search. If not, return to step S52 to continue expanding.
[0035] S56. Mark the current parent node as part of the path and add it to setclosedlist;
[0036] S57. Backtrack the nodes in setclosedlist, integrate them to obtain the planned path, and complete the path generation process.
[0037] The beneficial effects of this invention are:
[0038] 1. The present invention provides a full-coverage path planning method for a leveling robot, with local area coverage algorithm optimization: the PMTS algorithm (Prioritize Moving Towards Starting) is introduced, defining eight movement directions (N, S, E, W, NE, NW, SE, SW) and setting three-layer priority rules to guide the robot to move preferentially towards the starting point and horizontal direction, thereby efficiently bypassing obstacles and reducing missed areas.
[0039] 2. During the partial coverage process, this invention dynamically generates a list of target points for uncovered edge grids and uses a structural filtering mechanism to determine the status of neighboring grids and remove redundant points. It then selects the optimal connecting target point based on Euclidean distance.
[0040] 3. This invention uses an improved A* algorithm to guide the robot to move between sub-regions, optimizes the heuristic function calculation method, combines Manhattan and Euclidean distances, and introduces a dynamic weight adjustment strategy to control the search depth and path smoothness.
[0041] 4. By repeatedly calling the local coverage and region connection algorithms, the robot starts from the starting point and continues until all regions are completely covered, and outputs the complete path trajectory.
[0042] 5. This invention has fewer sub-region divisions, smoother coverage paths, fewer turning points, and significantly improved path search efficiency. At the same time, it significantly improves coverage completeness and first-coverage ratio. Attached Figure Description
[0043] Figure 1 This is a flowchart of a full-coverage path planning method for a leveling robot according to the present invention;
[0044] Figure 2 This is a flowchart of the PMTS algorithm of this invention;
[0045] Figure 3 This is a simulation comparison diagram of the PMTS algorithm and the BCD algorithm in this invention;
[0046] Figure 4 This refers to the target point selection principle in this invention;
[0047] Figure 5 This is a schematic diagram of the grid markings in this invention;
[0048] Figure 6 This is a flowchart of the A* algorithm in this invention;
[0049] Figure 7 The diagram illustrates the method for calculating the heuristic function distance in the optimized A* algorithm of this invention.
[0050] Figure 8 This is a comparison chart of simulation results between the A* algorithm and the improved A* algorithm in this invention;
[0051] Figure 9 The simulation results of the full-coverage path planning method are shown in the figure. Detailed Implementation
[0052] The present invention will be further described below with reference to the accompanying drawings and specific embodiments:
[0053] like Figures 1 to 8 As shown, the present invention provides a full-coverage path planning method for a leveling robot, comprising the following steps:
[0054] S1. Obtain an environmental raster map containing impassable areas.
[0055] In this embodiment, the spatial grid model M of the current working environment is loaded, and relevant variables are initialized. These variables include the starting point q_start, the leveling robot step size ε, the grid size l, the target point list target_points, and the Full Coverage Path used to store the final trajectory. The starting point is set to the current position current_pos, ready to begin the coverage task.
[0056] S2. In the current sub-region, call the priority starting point movement local path planning algorithm, namely the PMTS algorithm, to guide the robot to complete the full coverage operation in the region until it reaches the dead point of the current sub-region.
[0057] In step S2, a dead point refers to a location where the eight grids surrounding the robot's position are covered grids, obstacle grids, or boundaries, and can be identified as a dead point location.
[0058] The PMTS algorithm in step S2 involves eight movement directions: east, south, west, north, southeast, northeast, southwest, and northwest. Three priority rules are established for each direction. The first rule states that east, south, west, and north have higher priority than southeast, northeast, southwest, and northwest. The second rule states that if multiple feasible directions exist during path planning, the direction towards the starting position should be prioritized to effectively reduce path redundancy. The third rule further clarifies that horizontal directions have higher priority than vertical directions in determining specific directions, which helps improve the overall efficiency and consistency of path planning. In this embodiment, the horizontal direction is east or west, and the vertical direction is north or south.
[0059] In the local path planning process of step S2, the robot scans the grid status in eight directions around it in real time, identifies uncovered and passable grid cells, and adds them to the target point candidate list. Candidate target points must be located on the edge or boundary grid of an uncovered area and meet the predefined conditions for edge grids. The predefined conditions in step S2 refer to the following: if the left, right, front, back, and diagonally adjacent grid cells of a node are simultaneously in a free and uncovered state, then that node can be considered as not possessing edge / corner characteristics. Among the retained candidate target points, the target point with the smallest Euclidean distance to the current dead point is selected as the target position for the next connecting path, balancing path optimization and computational efficiency.
[0060] The PMTS algorithm in step S2 includes the following sub-steps:
[0061] S21. Initialize environment and parameters: Load the environment space model M for the leveling robot, and initialize the current position current_pos, grid size l, step size ε, and local coverage path.
[0062] S22. Environmental Information Acquisition: Use sensors to acquire current environmental data, update the grid status and synchronously update the environmental spatial model M to ensure a true reflection of the environment and avoid path deviation or coverage omissions caused by lag in environmental information.
[0063] S23. Determine movable directions: Determine whether movement is possible based on the grid status. Uncovered grids are considered movable directions, while covered or obstructed grids are considered infeasible directions. Using the information obtained in step S22, determine all movable directions for the current position.
[0064] S24. Check feasible paths: Check if there is a direction that can be moved in. If not, skip to step S27; if there is a feasible direction, proceed to the next step.
[0065] S25. Priority Direction Selection: According to the priority rules of the PMTS algorithm, the optimal direction is selected from all feasible directions, with priority given to the direction towards the starting point and the horizontal direction.
[0066] S26. Perform movement: The leveling robot moves once according to the selected direction and step size. After completion, it returns to step S22 to continue updating environmental information and making judgments.
[0067] S27. Complete Local Coverage: When the leveling robot reaches the end of the current area or a position where it can no longer move forward, the local coverage task ends and the path result Local Coverage Path is output.
[0068] Simulation results of PMTS and BCD algorithms are as follows: Figure 3 As shown, Figure 3 (a) shows the simulation results of the PMTS algorithm, and (b) shows the simulation results of the BCD algorithm.
[0069] In this embodiment, the PMTS local path planning algorithm is activated. While guiding the robot to complete coverage in the current sub-region, the sensors continuously collect information about the surrounding environment and update the spatial model.
[0070] S3. During the robot's movement, the robot checks the status of adjacent grid cells. If the coverage edge condition is met, the grid cells are added to the target point list.
[0071] Once the robot reaches the dead point in the area, the local path planning phase is complete.
[0072] S4. Eliminate unreasonable or redundant target points based on preset screening criteria to improve the efficiency of subsequent path generation.
[0073] The selection principle of the target point selection mechanism in step S4 is as follows:
[0074] A target point list is established. While performing local area coverage, the robot scans the grid status in eight directions, selecting uncovered and passable grid cells and recording them in the target point list as candidate points for subsequent path planning. If a target point has been covered by the robot, it should be immediately removed from the list to ensure data accuracy. Simultaneously, target points selected in area connection path planning should also be removed from the list to optimize the distribution of target points.
[0075] In step S4, removing unreasonable or redundant target points constitutes the target point selection. Target points should be selected from the edge positions of uncovered areas. If the left, right, front, back, or diagonally adjacent grid cells of a node are simultaneously in a free and uncovered state, then that node can be considered to lack edge / corner characteristics. Figure 4 As shown.
[0076]
[0077] ρ(m)=μ(d1,d5)+μ(d2,c6)+μ(d3,d7)+μ(d4,d8)
[0078] Wherein, μ(d) is defined i ,d j This describes the state relationships between these grids, where i and j range from 1 to 8, m is the grid the robot is currently in, and d... i and d j Number the eight grid cells surrounding the robot, such as... Figure 5 As shown, a discriminant function ρ(m) will be introduced to determine whether grid m meets the selection criteria for edge target points.
[0079] When ρ(m) ≥ 1, it indicates that the node is located in a continuous free region and does not have boundary features. Therefore, it is determined to be an invalid target point and should be removed from the candidate set. When ρ(m) = 0, it means that there are no two simultaneously free and uncovered grids in its horizontal, vertical, or diagonal directions. In this case, the grid is located at a corner position and can be retained as a reference target point for subsequent path planning.
[0080] The target point is chosen by selecting the position with the smallest Euclidean distance from the current dead point, which will be the endpoint of the next connecting path.
[0081] S5. Select the target point with the closest Euclidean distance to the current dead point as the entry point of the next sub-region, and call the improved A* algorithm of the region connection path planning algorithm to generate the connection path.
[0082] The region connectivity path planning algorithm in step S5 is a weighted average heuristic function method that combines Manhattan distance and Euclidean distance. Traditional A* algorithms expand some redundant nodes during path search. This phenomenon stems from the fact that the heuristic function h^*(n) cannot be flexibly adjusted according to the real-time search process. When the estimated cost is lower than the actual cost, the number of nodes expanded by the algorithm increases significantly, leading to a decrease in search efficiency. In this case, the weight of h^*(n) needs to be increased to speed up the calculation. When the estimated cost h^*(n) equals the actual cost, only necessary nodes are traversed. If the estimated cost is higher than the actual cost, although the search speed is improved, the optimal path is easily missed. Therefore, the weight should be reduced to balance path quality and efficiency.
[0083] The principle of the A* algorithm: The A* algorithm introduces two key cost quantities: one is the actual cost from the current point n to the starting point, and the other is the estimated cost from the current point n to the target point. The selection method for the next child node is determined by summing these two costs. The A* algorithm evaluation function is calculated as follows:
[0084] f(n) = g(n) + h(n)
[0085] In the formula, g(n) is the known shortest path from the starting node to node n; h(n) is a heuristic estimate of the cost from node n to the target node.
[0086] like Figure 5 As shown, the specific steps of the A* algorithm in step S5 are as follows:
[0087] S51. Set the starting point as S and the ending point as G; initialize a list named setopenlist to store candidate nodes that have not yet been expanded, and create setclosedlist to record nodes that have been visited; the starting node S will be added to setopenlist first, while setclosedlist is still empty at this time, making preliminary preparations for the path search process.
[0088] S52. Expand the current parent node, add all its reachable adjacent child nodes to setopenlist, and move the parent node itself into setclosedlist, indicating that the node has been visited.
[0089] S53. Check if setopenlist is empty. If setopenlist is empty, it means that no feasible path can be found during the search process, and the program terminates; otherwise, continue to the next step.
[0090] S54. Calculate the cost value of the current expandable child node according to the A* evaluation function, and select the child node with the smallest evaluation value as the new parent node.
[0091] S55. Determine whether the currently selected parent node is the target point G. If it is, return to the starting point from the target point by backtracking to generate the final path and end the search. If not, return to step S52 to continue expanding.
[0092] S56. Mark the current parent node as part of the path and add it to setclosedlist.
[0093] S57. Backtrack the nodes in setclosedlist, integrate them to obtain the planned path, and complete the path generation process.
[0094] like Figure 6 As shown, the bend in the middle of the triangle represents the improved heuristic function path. Figure 6 In triangle ABC, the angle bisectors of ∠BAC and ∠BCA intersect at point X. The dashed line segments OX, HX, and HI are of equal length. According to the Pythagorean theorem and the angle bisector theorem, we can obtain:
[0095]
[0096] The improved heuristic function calculation expression is:
[0097]
[0098] When the estimated cost is lower than the actual cost, the number of nodes expanded by the algorithm increases significantly, leading to a decrease in search efficiency. In this case, it is necessary to appropriately increase h. * The weight values of (n) are used to speed up the calculation. When the estimated cost h * When (n) equals the actual cost, the search path is optimal, traversing only necessary nodes. If the estimated cost is higher than the actual cost, although the search speed improves, the optimal path is easily missed. Therefore, the weights should be appropriately reduced to balance path quality and efficiency. The improved evaluation function is calculated as follows:
[0099]
[0100] In the formula: f(n) is the total cost from the starting point to the target point; g(n) is the actual cost from the current node to the starting point; h * (n) is the estimated cost from the current node to the target node; d is the distance between the current node and the target node; D is the distance from the starting point to the target node.
[0101] Simulation results of A* and improved A* algorithms are as follows: Figure 7 As shown, Figure 7(a) shows the simulation result of the A* algorithm, and (b) shows the simulation result of the improved A* algorithm. In this embodiment, the improved A* algorithm, which uses a region connection path planning algorithm to generate connection paths, optimizes the calculation method of the heuristic function. By introducing dynamic weight coefficients, the computational complexity and running time are reduced, thereby improving the efficiency of the leveling robot operation.
[0102] S6. Proceed along the planned path to the target point, update the current location to that point, restart the partial coverage process, and repeat the process.
[0103] S7. If the current target point list is empty, it means that all reachable areas have been visited, and the coverage is complete.
[0104] In this embodiment, if the current target point list `target_points` is empty, it means that all reachable areas have been visited, and the program jumps to the next step, which is to output the robot's entire movement trajectory as a Full Coverage Path, thus ending the full coverage path planning. Otherwise, proceed to step S5.
[0105] The principle of this invention is as follows: On a map containing obstacles, the PMTS algorithm is used for coverage until a dead point is encountered. A target point selection mechanism is then used to select the most suitable target point located in the uncovered area, and an improved A* algorithm is initiated to derive the optimal path between the dead point and the target point. After reaching the uncovered area, the PMTS algorithm is used again for coverage until full coverage of the environmental space is achieved.
[0106] The core objective of the full-coverage path planning algorithm is to plan an optimal path for the robot, starting from its initial position, that covers all passable areas in the environment while minimizing the total travel distance. Simulation results are as follows: Figure 8 As shown.
[0107] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of this invention.
Claims
1. A method for full-coverage path planning for a leveling robot, characterized in that, Includes the following steps: S1. Obtain an environmental raster map containing impassable areas; Load the spatial grid model of the current working environment Initialize relevant variables, including the starting point q_start and the leveling robot step size. Grid size The system includes a list of target points (target_points) and a Full Coverage Path for storing the final trajectory; the starting point is set to the current position (current_pos), ready to begin the coverage task. S2. In the current sub-region, call the priority starting point movement local path planning algorithm, namely the PMTS algorithm, to guide the robot to complete the full coverage operation in the region until it reaches the dead point of the current sub-region. The PMTS algorithm in step S2 includes the following sub-steps: S21. Initialize the environment and parameters: Load the environment space model for the leveling robot. Initialize the current position (current_pos) and the grid size. Step length And local coverage path; S22. Environmental Information Acquisition: Acquire current environmental data using sensors, update the raster status, and synchronously update the environmental spatial model. This ensures an accurate reflection of the environment and avoids path deviations or omissions due to outdated environmental information. S23. Determine the movable direction: Determine whether movement is possible based on the grid status. Uncovered grids are considered feasible directions, while covered or obstacle-filled grids are considered infeasible directions. Using the information obtained in step S22, identify all movable directions for the current position. S24. Detect feasible paths: Check if there is a direction that can be moved; if not, jump to step S27. If a feasible direction exists, proceed to the next step; S25. Priority Direction Selection: According to the priority rules of the PMTS algorithm, the optimal direction is selected from all feasible directions, giving priority to the direction towards the starting point and the horizontal direction. S26. Perform movement: The leveling robot moves once according to the selected direction and step size. After completion, it returns to step S22 to continue updating environmental information and making judgments. S27. Complete Local Coverage: When the leveling robot reaches the end of the current area or a position where it can no longer move forward, the local coverage task ends and the path result Local Coverage Path is output. S3. During the robot's movement, the robot checks the status of adjacent grid cells. If the coverage edge condition is met, the grid cells are added to the target point list. S4. Eliminate unreasonable or redundant target points based on preset screening criteria to improve the efficiency of subsequent path generation. S5. Select the target point with the closest Euclidean distance to the current dead point as the entry point of the next sub-region, and call the improved A* algorithm of the region connection path planning algorithm to generate the connection path; S6. Proceed along the planned path to the target point, update the current location to that point, restart the partial coverage process, and repeat the process. S7. If the current target point list is empty, it means that all reachable areas have been visited, and the coverage is complete. The PMTS algorithm in S2 involves eight movement directions: east, south, west, north, southeast, northeast, southwest, and northwest. It also sets three priority rules for these directions. The first rule states that east, south, west, and north have higher priority than southeast, northeast, southwest, and northwest. The second rule states that if multiple feasible directions exist during path planning, the direction towards the starting position should be selected first to effectively reduce path redundancy. The third rule further clarifies that horizontal directions have higher priority than vertical directions in determining specific directions.
2. The method for full-coverage path planning of a leveling robot according to claim 1, characterized in that: During the local path planning process in S2, the robot scans the grid status in eight directions around it in real time, identifies uncovered grid cells that are passable, and adds them to the target point candidate list. Candidate target points must be located on the edge or boundary grid of an uncovered area and meet the predefined conditions of edge grids. If all grids surrounding a candidate point are covered areas or obstacles, it is considered to lack edge attributes and is removed to avoid redundant coverage and path breaks. Among the retained candidate target points, the target point with the smallest Euclidean distance to the current dead point is selected as the target position for the next connecting path.
3. The method for full-coverage path planning of a leveling robot according to claim 1, characterized in that: The region connectivity path planning algorithm in S5 is a weighted average heuristic function method that combines Manhattan distance and Euclidean distance. Traditional A* algorithms expand some redundant nodes during path search. This phenomenon stems from the fact that the heuristic function h^*(n) cannot be flexibly adjusted according to the real-time search process. When the estimated cost is lower than the actual cost, the number of nodes expanded by the algorithm increases significantly, leading to a decrease in search efficiency. In this case, the weight value of h^*(n) needs to be increased to speed up the calculation. When the estimated cost h^*(n) equals the actual cost, only necessary nodes are traversed. If the estimated cost is higher than the actual cost, although the search speed is improved, the optimal path is easily missed.
4. The method for full-coverage path planning of a leveling robot according to claim 1, characterized in that: The selection principle of the target point selection mechanism in S4 is as follows: a target point list is established. While covering a local area, the robot will scan the grid status in eight directions around it, select grid cells that have not yet been covered and are in a passable state, and record them in the target point list as candidate points for subsequent path planning. If a target point has been covered by the robot, it should be removed from the list immediately to ensure data accuracy; at the same time, target points selected in the area connection path planning should also be removed from the list to optimize the distribution of target points.
5. The method for full-coverage path planning of a leveling robot according to claim 1, characterized in that: In S4, removing unreasonable or redundant target points is the screening of target points. Target points should be selected from the edge positions of areas that are not yet covered. If the left, right, front, back, or diagonally adjacent grids of a node are all in a free and uncovered state, then the node can be considered to lack edge and corner characteristics. ； ； When the condition is met, it indicates that the node is located in a continuous free region and does not possess boundary features; therefore, it is determined to be an invalid target point and should be removed from the candidate set. When there are no two simultaneously uncovered grids in its horizontal, vertical, or diagonal directions, the grid is located at a corner position and can be retained as a reference target point for subsequent path planning.
6. The method for full-coverage path planning of a leveling robot according to claim 1, characterized in that: The specific steps of the A* algorithm in S5 are as follows: S51. Set the starting point as S and the ending point as G; initialize a list named setopenlist to store candidate nodes that have not yet been expanded, and create setclosedlist to record nodes that have been visited; the starting node S will be added to setopenlist first, while setclosedlist is still empty at this time, making preliminary preparations for the path search process; S52. Expand the current parent node, add all its reachable adjacent child nodes to setopenlist, and move the parent node itself into setclosedlist, indicating that the node has been visited; S53. Check if setopenlist is empty. If setopenlist is empty, it means that no feasible path could be found during the search, and the program terminates; otherwise, continue to the next step. S54. Calculate the cost value of the current expandable child node according to the A* evaluation function, and select the child node with the smallest evaluation value as the new parent node. S55. Determine whether the currently selected parent node is the target point G. If it is, return to the starting point from the target point by backtracking to generate the final path and end the search. If not, return to step S52 to continue expanding. S56. Mark the current parent node as part of the path and add it to setclosedlist; S57. Backtrack the nodes in setclosedlist, integrate them to obtain the planned path, and complete the path generation process.