Path planning method and device for photovoltaic unmanned cleaning vehicle and storage medium

Abstract

This invention relates to the field of path planning technology, specifically disclosing a path planning method, device, and storage medium for a photovoltaic unmanned cleaning vehicle. The method includes: analyzing a preset grid map of the photovoltaic panel area; performing offline path planning on the photovoltaic panel area based on an offline path search algorithm to obtain preliminary results; wherein the evaluation function of the offline path search algorithm includes at least a heuristic function and a path cost function, the heuristic function being a cost function determined based on the target point heuristic value information of each cleaning operation target point, and the path cost function being a cost function determined based on the kinematic characteristics and path movement consumption of the photovoltaic unmanned cleaning vehicle; and optimizing the preliminary results of the offline path planning according to preset tracking rules of the photovoltaic unmanned cleaning vehicle. The path planning method for photovoltaic unmanned cleaning vehicles provided by this invention can effectively improve the path planning capability of photovoltaic unmanned cleaning vehicles while improving the cleaning effect.

Description

Technical Field

[0001] This invention relates to the field of path planning technology, and in particular to a path planning method, a path planning device, and a storage medium for a photovoltaic unmanned cleaning vehicle. Background Technology

[0002] Photovoltaic power generation is a significant source of clean energy; however, dust on the surface of photovoltaic panels can greatly affect their power generation efficiency, making photovoltaic panel cleaning a pressing issue. In the context of industrial automation, unmanned vehicles (UAVs) offer a fast and effective solution for cleaning photovoltaic panels. The large area of ​​photovoltaic power plants, the unstructured roads along the driving paths, and the non-unique layout of photovoltaic panels pose significant challenges and time costs for manually collecting UAV paths. Therefore, automated path planning to ensure effective cleaning is essential. Unlike path planning for ordinary autonomous vehicles, photovoltaic panels are both obstacles and the cleaning vehicle's work area. This necessitates path planning that not only avoids obstacles but also traverses them while ensuring the optimal path. Existing autonomous driving path planning relies on perception and localization technologies, requiring expensive sensors and powerful hardware for real-time online calculations. This method amplifies errors in various stages, leading to incomplete cleaning of photovoltaic panels and blind spots. Furthermore, its focus is on finding a collision-free optimal (usually shortest) path, lacking obstacle traversal, which does not meet the requirements for photovoltaic panel cleaning.

[0003] Existing technologies have proposed various path planning methods for photovoltaic panel cleaning. One method, for example, involves a path planning approach for an intelligent cleaning robot for photovoltaic arrays. This method divides the photovoltaic array into a two-dimensional index array, determines the starting point of the operation based on the distribution of cleaned and uncleaned portions, and plans a "bow"-shaped path. It then constructs an undirected graph based on the connectivity of road points, wall points, and parking spots in the scene, and calculates the shortest path from the robot's current point to the starting point using Dijkstra's algorithm. However, this method has drawbacks. Forcing a fixed connection order and defining the cleaning path as a "bow" shape makes it difficult to cover all areas of a large photovoltaic field. Furthermore, it requires the construction of an undirected graph based on scene points, and the accuracy of the graph needs verification. Much of this work is manually defined, and it does not support automatic path planning for the entire photovoltaic scene. Another approach is a path navigation system for photovoltaic cleaning robots. This system includes multiple modules such as perception, positioning, path planning, obstacle avoidance, and control. It integrates numerous sensors, including cameras, lidar, infrared, and meteorological sensors, and achieves fully automated driving through online real-time calculations, completing both the photovoltaic cleaning task and the scheduling of the cleaning vehicle. The drawbacks of this method are that the perception, localization, and obstacle avoidance modules of the fully automated cleaning process heavily rely on information from multiple sensors, and the online computing requires powerful computer hardware support to meet real-time requirements, resulting in high equipment costs. Furthermore, this method focuses on the system design level and does not propose specific operational logic for the path planning module in photovoltaic scenarios.

[0004] Therefore, how to propose a solution that can effectively improve the path planning capability of photovoltaic unmanned cleaning vehicles and improve the cleaning effect, taking into account the characteristics of unstructured multi-objective path planning, relatively fixed maps, non-unique arrangement of photovoltaic panels, large working area, and real-time cleaning blind spots in photovoltaic unmanned cleaning scenarios, has become a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] This invention provides a path planning method, a path planning device, and a storage medium for photovoltaic unmanned cleaning vehicles, solving the problem of lacking path planning for cleaning vehicles in photovoltaic unmanned cleaning scenarios in related technologies.

[0006] As a first aspect of the present invention, a path planning method for a photovoltaic unmanned cleaning vehicle is provided, comprising:

[0007] The preset grid map of the photovoltaic panel area is analyzed to determine the target points for cleaning operations. Each target point for cleaning operations is configured with at least the target point coordinate location information, target point access status information, target point cleaning status information, and target point heuristic value information used to guide the cleaning sequence.

[0008] The task start and end points of the photovoltaic panel area are determined according to the preset path planning task, and offline path planning is performed on the photovoltaic panel area based on the offline path search algorithm until the access ratio of the cleaning operation target point reaches the preset ratio threshold, and the preliminary results of offline path planning are obtained. The evaluation function of the offline path search algorithm includes at least a heuristic function and a path cost function. The heuristic function is a cost function determined based on the target point heuristic value information of each cleaning operation target point, and the path cost function is a cost function determined based on the kinematic characteristics and path movement consumption of the photovoltaic unmanned cleaning vehicle.

[0009] The preliminary offline path planning results are optimized based on the preset tracking rules of the photovoltaic unmanned cleaning vehicle to obtain the path planning results of the photovoltaic unmanned cleaning vehicle.

[0010] Furthermore, the preset grid map of the photovoltaic panel area is analyzed, including:

[0011] The preset grid map is traversed and searched using a graph search algorithm to locate the positions of all photovoltaic panels in the photovoltaic panel area;

[0012] The boundary line information of each photovoltaic panel is determined based on the position of each photovoltaic panel and the orientation information of the preset grid map. The boundary line information of each photovoltaic panel includes at least one boundary line along the length direction of the photovoltaic panel.

[0013] Based on the boundary information of each photovoltaic panel and the preset interval distance, multiple cleaning operation target points corresponding to each photovoltaic panel are determined.

[0014] Furthermore, based on the boundary line information of each photovoltaic panel and the preset interval distance, multiple cleaning operation target points corresponding to each photovoltaic panel are determined, including:

[0015] The preset interval distance is determined based on the current dimensions of the photovoltaic unmanned cleaning vehicle and the preset cleaning range of the cleaning unit;

[0016] The locations of multiple cleaning target points for each photovoltaic panel are determined based on the boundary line information of each photovoltaic panel and the preset interval distance.

[0017] Further, the start and end points of the task in the photovoltaic panel area are determined according to the preset path planning task, and offline path planning is performed on the photovoltaic panel area based on the offline path search algorithm until the access ratio of the cleaning operation target point reaches a preset ratio threshold, thereby obtaining preliminary results of offline path planning, including:

[0018] The starting point for the unmanned photovoltaic cleaning vehicle to enter the photovoltaic panel area and the ending point for leaving the photovoltaic panel area are determined according to the preset path planning task.

[0019] The evaluation function of the offline path search algorithm is determined based on the target point heuristic value of each cleaning operation target point, the kinematic characteristics of the photovoltaic unmanned cleaning vehicle, and the path movement consumption.

[0020] Offline path planning is performed on the photovoltaic panel area based on the offline path search algorithm, target point coordinate location information, target point access status information, and target point cleaning status information.

[0021] Determine whether the access ratio of the current cleaning operation target point has reached the preset ratio threshold;

[0022] If the access ratio of the current cleaning operation target point reaches the preset ratio threshold, the search will stop and the preliminary results of offline path planning will be obtained.

[0023] Furthermore, the evaluation function of the offline path search algorithm is determined based on the target point heuristic value of each cleaning operation target point, the kinematic characteristics of the photovoltaic unmanned cleaning vehicle, and the path movement consumption, including:

[0024] The heuristic cost function is determined based on the target point heuristic value of each cleaning operation target point and the distance between the current grid point of the photovoltaic unmanned cleaning vehicle and the cleaning operation target point;

[0025] The path length cost, turning cost, and reversing cost are determined based on the kinematic characteristics of the photovoltaic unmanned cleaning vehicle, and the path cost function is determined based on the path length cost, turning cost, and reversing cost.

[0026] The evaluation function of the offline path search algorithm is determined based on the heuristic cost function and the path cost function.

[0027] Furthermore, offline path planning is performed on the photovoltaic panel area based on the offline path search algorithm, target point coordinate location information, target point access status information, and target point cleaning status information, including:

[0028] The cleaning target points within the photovoltaic panel area are accessed according to the offline path search algorithm.

[0029] When passing through a grid occupied by a cleaning operation target point, the target point access status information and target point cleaning status information of that grid are updated, and the access ratio of the current cleaning operation target point is calculated based on the target point access status information.

[0030] Furthermore, the preliminary results of the offline path planning are optimized according to the preset tracking rules of the photovoltaic unmanned cleaning vehicle, including:

[0031] The preset tracking rules for the photovoltaic unmanned cleaning vehicle are determined based on the curvature limit, the uniformity of curvature change, the uniformity of spacing, and the probability of collision.

[0032] Based on the preset tracking rules, the target points for abnormal cleaning operations are identified in the offline path planning results.

[0033] The offline path after removing abnormal cleaning operation target points is optimized to obtain the path planning results of the photovoltaic unmanned cleaning vehicle.

[0034] Furthermore, the offline path after removing abnormal cleaning operation target points is optimized, including:

[0035] The offline path after removing abnormal cleaning target points is smoothed using the gradient descent method to increase smoothness.

[0036] The offline path with increased smoothness is densified using polynomial interpolation to obtain the path planning results for the photovoltaic unmanned cleaning vehicle.

[0037] In another aspect, a path planning device for a photovoltaic unmanned cleaning vehicle is provided, for implementing the path planning method for a photovoltaic unmanned cleaning vehicle described above, wherein the path planning device for the photovoltaic unmanned cleaning vehicle includes:

[0038] The grid map analysis module is used to analyze the preset grid map of the photovoltaic panel area to determine the target points for cleaning operations. Each target point for cleaning operations is configured with at least the target point coordinate location information, target point access status information, target point cleaning status information, and target point heuristic value information used to guide the cleaning sequence.

[0039] The multi-objective traversal path planning module is used to determine the start and end points of the photovoltaic panel area according to the preset path planning task, and to perform offline path planning on the photovoltaic panel area based on the offline path search algorithm until the access ratio of the cleaning operation target point reaches the preset ratio threshold, and obtain the preliminary results of the offline path planning. The evaluation function of the offline path search algorithm includes at least a heuristic function and a path cost function. The heuristic function is a cost function determined based on the target point heuristic value information of each cleaning operation target point, and the path cost function is a cost function determined based on the kinematic characteristics and path movement consumption of the photovoltaic unmanned cleaning vehicle.

[0040] The path post-processing module is used to optimize the preliminary offline path planning results according to the preset tracking rules of the photovoltaic unmanned cleaning vehicle, so as to obtain the path planning result of the photovoltaic unmanned cleaning vehicle.

[0041] As another aspect of the present invention, a storage medium is provided, wherein computer instructions are stored, which, when loaded and executed by a processor, implement the path planning method for photovoltaic unmanned cleaning vehicles described above.

[0042] The path planning method for unmanned photovoltaic cleaning vehicles provided by this invention analyzes a preset grid map, traverses and extracts map information, and redefines the grid information of multiple target points. Based on the kinematic characteristics of the unmanned photovoltaic cleaning vehicle, path movement costs, and target point heuristics, a cost function is constructed. Then, offline path planning is performed offline using an offline path search algorithm and the results of offline path planning analysis. This achieves automatic path planning for multi-target traversal across the entire photovoltaic power plant scenario. This path planning method for unmanned photovoltaic cleaning vehicles is based on a preset grid map and is performed offline, eliminating the need for real-time monitoring. Therefore, it can reduce accumulated errors and overcome cleaning blind spots, effectively improving the path planning capability of the unmanned photovoltaic cleaning vehicle while simultaneously enhancing the cleaning effect. Attached Figure Description

[0043] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the following detailed description to explain the invention, but do not constitute a limitation thereof.

[0044] Figure 1 A flowchart of the path planning method for photovoltaic unmanned cleaning vehicles provided by the present invention.

[0045] Figure 2 This is a flowchart illustrating the analysis of a preset grid map of a photovoltaic panel area, as provided by the present invention.

[0046] Figure 3 This is a top view of the photovoltaic panel area with the distance between the target point and the edge line provided by the present invention.

[0047] Figure 4 This is a flowchart for obtaining preliminary results of offline path planning provided by the present invention.

[0048] Figure 5a This is a schematic diagram of node expansion for the traditional A* algorithm provided by the present invention.

[0049] Figure 5b This is a schematic diagram of node expansion for the Hybrid-A* algorithm provided by the present invention.

[0050] Figure 6 This is a schematic diagram of the Hybrid-A* child node definition method provided by the present invention.

[0051] Figure 7 A schematic diagram illustrating the generation of the blind zone provided by this invention.

[0052] Figure 8 This is a flowchart for optimizing the preliminary results of offline path planning, as provided by the present invention.

[0053] Figure 9 This is a structural block diagram of the path planning device for photovoltaic unmanned cleaning vehicles provided by the present invention.

[0054] Figure 10 This is a structural block diagram of the electronic device provided by the present invention. Detailed Implementation

[0055] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0056] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0057] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of the invention described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0058] This embodiment provides a path planning method for photovoltaic unmanned cleaning vehicles. Figure 1 This is a flowchart of a path planning method for a photovoltaic unmanned cleaning vehicle provided according to an embodiment of the present invention, such as... Figure 1 As shown, it includes:

[0059] S100. Analyze the preset grid map of the photovoltaic panel area to determine the target points for cleaning operations. Each target point for cleaning operations is configured with at least the target point coordinate location information, the target point access status information, the target point cleaning status information, and the target point heuristic value information used to guide the cleaning sequence.

[0060] In this embodiment of the invention, each photovoltaic power plant can generate its own grid map, so the preset grid map of the photovoltaic power plant can be directly imported and analyzed.

[0061] Specifically, in this embodiment of the invention, the preset grid map can be traversed using the Dijkstra graph search algorithm, and all photovoltaic panels can be identified based on the shape and area of ​​the obstacles. Then, the boundary line of the photovoltaic panel (reference edge line of the photovoltaic unmanned cleaning vehicle) can be identified based on the map direction and the photovoltaic panel position information. At a certain distance from the boundary line of the photovoltaic panel, the same number of target points are defined along the boundary line of the photovoltaic panel. The grid occupied by the target point is defined with a fixed heuristic value, an access flag (all initialized to unvisited), and a cleaning work flag (all initialized to uncleaned).

[0062] S200. Determine the start and end points of the photovoltaic panel area according to the preset path planning task, and perform offline path planning for the photovoltaic panel area based on the offline path search algorithm until the access ratio of the cleaning operation target point reaches the preset ratio threshold, and obtain the preliminary results of the offline path planning. The evaluation function of the offline path search algorithm includes at least a heuristic function and a path cost function. The heuristic function is a cost function determined based on the target point heuristic value information of each cleaning operation target point, and the path cost function is a cost function determined based on the kinematic characteristics and path movement consumption of the photovoltaic unmanned cleaning vehicle.

[0063] In this embodiment of the invention, after defining the start and end points of the task, the Hybrid-A* algorithm, which considers grid heuristics, kinematic characteristics of the cleaning vehicle, and path movement consumption, is used for offline path search. During the search process, visited and cleaned (planned path passed through) target points are recorded. The cleaning task end point can only be visited when the target point visit ratio reaches a predetermined requirement.

[0064] S300. Optimize the preliminary offline path planning results according to the preset tracking rules of the photovoltaic unmanned cleaning vehicle to obtain the path planning results of the photovoltaic unmanned cleaning vehicle.

[0065] In this embodiment of the invention, based on the preliminary results of the offline path planning obtained above, the curvature, spacing, and other points that do not meet the tracking requirements of the unmanned vehicle can be optimized according to the preset tracking rules of the photovoltaic unmanned cleaning vehicle, and finally a path that can be directly provided to the unmanned vehicle for tracking can be formed.

[0066] Therefore, the path planning method for photovoltaic unmanned cleaning vehicles provided by this invention analyzes a preset grid map, traverses and extracts map information, and redefines the grid information of multiple target points. It then constructs a cost function based on the kinematic characteristics of the photovoltaic unmanned cleaning vehicle, path movement costs, and target point heuristics. Finally, it combines the results of offline path planning analysis with an offline path search algorithm to perform offline multi-target traversal path planning. This achieves automatic path planning for multi-target traversal across the entire photovoltaic power plant scenario. This path planning method for photovoltaic unmanned cleaning vehicles is based on a preset grid map and is performed offline, eliminating the need for real-time monitoring. Therefore, it can reduce accumulated errors and overcome cleaning blind spots, effectively improving the path planning capability of photovoltaic unmanned cleaning vehicles while simultaneously enhancing cleaning performance.

[0067] As a specific embodiment of the present invention, a preset grid map of the photovoltaic panel area is analyzed, such as... Figure 2 As shown, it includes:

[0068] S110. The preset grid map is traversed and searched according to the graph search algorithm to locate the positions of all photovoltaic panels in the photovoltaic panel area;

[0069] In this embodiment of the invention, specifically for a known photovoltaic field grid map, the Dijkstra graph search algorithm can be used to filter all obstacles in the grid map based on the actual length, width range and layout of the photovoltaic panels, thereby locating all the photovoltaic panels and determining the position coordinate information of each photovoltaic panel on the grid map.

[0070] S120. Determine the boundary line information of each photovoltaic panel based on the position of each photovoltaic panel and the direction information of the preset grid map. The boundary line information of each photovoltaic panel includes at least one boundary line along the length direction of the photovoltaic panel.

[0071] In this embodiment of the invention, after determining the position of each photovoltaic panel, the boundary line information of each photovoltaic panel can be determined based on the orientation of a preset grid map. Specifically, most photovoltaic panels are oriented south to better receive sunlight, therefore, this can be combined with... Figure 3 The top view of the photovoltaic panels shows that the southern boundary line of each photovoltaic panel is determined. This not only facilitates the cleaning operation of the cleaning vehicle, but also reduces blind spots and improves cleaning efficiency.

[0072] More specifically, based on the boundary line information of each photovoltaic panel and the preset interval distance, multiple cleaning operation target points corresponding to each photovoltaic panel are determined, including:

[0073] 1) Determine the preset interval distance based on the current dimensions of the photovoltaic unmanned cleaning vehicle and the preset cleaning range of the cleaning unit;

[0074] 2) Determine the location of multiple cleaning target points for each photovoltaic panel based on the boundary line information of each photovoltaic panel and the preset interval distance.

[0075] In this embodiment of the invention, based on the location of each photovoltaic panel and the map orientation, a target point with the same number of grid cells is defined along the south line at a distance K on one side of the south line. The distance K(x) (where x represents the number of grid cells) is defined as follows:

[0076] K(x) = F(x) + G(x),

[0077] Wherein, F(x) is defined based on the length, width, and height of the unmanned cleaning vehicle, and G(x) is defined based on the cleanable range of the cleaning components of the unmanned cleaning vehicle (robotic arm, nozzles, etc.). Therefore, the rationality of the definition of K(x) determines the cleaning effect of each photovoltaic panel and the effect of blind spot treatment. In this embodiment of the invention, a schematic diagram of the definition of distance K is shown below. Figure 3 As shown.

[0078] S130. Determine multiple cleaning target points corresponding to each photovoltaic panel based on the boundary line information of each photovoltaic panel and the preset interval distance.

[0079] It should be understood that after determining the grid used for the target points, a heuristic constant C, an access flag FLAG, and a cleanup flag CLEAN are added to all target point grids. The heuristic value C is associated with the heuristics of the Hybrid-A* algorithm, the access flag FLAG is associated with target point management, and the cleanup flag CLEAN is associated with the cleaning behavior of the unmanned cleaning vehicle.

[0080] Therefore, in this embodiment of the invention, the Dijkstra graph search algorithm is used to traverse the grid map, identifying all photovoltaic panels based on their shape, area, and layout. Then, the southern edge of the photovoltaic panels is identified using map orientation and panel location information. Target points of the same number of grid cells are defined along the southern edge of the photovoltaic panels at certain intervals. Each grid cell occupied by a target point has a fixed heuristic value, an access flag (visited and unvisited), and a cleaning flag. The heuristic value guides the subsequent path planning algorithm, the access flag updates the traversal information of the target points, and the cleaning flag is used by the cleaning module to perform cleaning operations on the photovoltaic panels.

[0081] As a specific embodiment of the present invention, the start and end points of the photovoltaic panel area are determined according to the preset path planning task, and offline path planning is performed on the photovoltaic panel area based on the offline path search algorithm until the access ratio of the cleaning operation target point reaches a preset ratio threshold, thereby obtaining preliminary results of offline path planning, such as... Figure 4 As shown, it includes:

[0082] S210. Determine the starting point for the unmanned photovoltaic cleaning vehicle to enter the photovoltaic panel area and the ending point for it to leave the photovoltaic panel area according to the preset path planning task.

[0083] In this embodiment of the invention, the start and end points of the task can be determined based on a preset path planning task, for example, such as... Figure 3 As shown, the top right corner can be used as the starting point of the task, and the bottom left corner as the ending point. The specific starting and ending points can be set as needed, and there are no restrictions here.

[0084] S220. Determine the evaluation function of the offline path search algorithm based on the target point heuristic value of each cleaning operation target point, the kinematic characteristics of the photovoltaic unmanned cleaning vehicle, and the path movement consumption.

[0085] In this embodiment of the invention, a heuristic cost function can be determined by combining the target point heuristic value of each cleaning operation target point, and a path cost function can be determined based on the kinematic characteristics of the photovoltaic unmanned cleaning vehicle and the path movement consumption. The evaluation function of the offline path search algorithm can be determined based on the heuristic cost function and the path cost function.

[0086] Specifically, the evaluation function of the offline path search algorithm is determined based on the target point heuristic value of each cleaning operation target point, the kinematic characteristics of the photovoltaic unmanned cleaning vehicle, and the path movement consumption, including:

[0087] 1) Determine the heuristic cost function based on the target point heuristic value of each cleaning operation target point and the distance between the current grid point of the photovoltaic unmanned cleaning vehicle and the cleaning operation target point;

[0088] It should be understood that the closer the heuristic function is to reality, the higher the efficiency of path search. Because existing heuristics only use Euclidean distance (the straight-line distance between two points) or Manhattan distance (the sum of the absolute axial distances of two points in the coordinate system) as heuristic functions, without considering obstacles, they cannot effectively guide path search in photovoltaic scenarios with a large number of orderly arranged photovoltaic panels. Based on this, this embodiment of the invention designs a heuristic based on the actual distance from the current grid point to the target point (cleaning point or task endpoint), considering obstacles, and incorporates a target point heuristic value constant C from the grid map redefinition module to ensure priority search towards the custom target point. Specifically, the heuristic function expression is as follows:

[0089] h(s) = d(s) + C,

[0090] Where d(s) represents the actual distance between the current grid position and the target point, which can be obtained by methods such as Dijkstra or RRT*. C represents the target point heuristic value, which is usually a constant and the optimal value of the constant can be determined experimentally.

[0091] 2) Determine the path length cost, turning cost, and reversing cost based on the kinematic characteristics of the photovoltaic unmanned cleaning vehicle, and determine the path cost function based on the path length cost, turning cost, and reversing cost;

[0092] In this embodiment of the invention, the path cost function g(s) is used to evaluate the actual cost of moving from one grid cell to another. To ensure that the planned path better reflects the actual photovoltaic cleaning scenario and avoids energy consumption caused by meaningless turns and reversing, the path cost function considers the cost of path length, turns, and reversing. The specific calculation formula is as follows:

[0093] g now =g parent +k1×g turn +k2×g back ,

[0094] Among them, g now G represents the cost of the current target point. parent g represents the cost of the parent node of the current target point. turn g represents the cost of turning. back The value represents the cost of reversing, and k1 and k2 both represent corresponding coefficients, which can be obtained experimentally.

[0095] In this embodiment of the invention, the definition of child nodes can be specifically understood as follows: the traditional A* path search algorithm expands child nodes in eight directions, can only access fixed positions within the grid, and does not consider vehicle kinematic parameters; therefore, the calculated path cannot be used directly. The Hybrid-A* algorithm considers the vehicle's minimum turning radius and front wheel steering angle, enabling access to any position within the grid, and the generated path meets vehicle kinematic requirements. The node expansion of the traditional A* algorithm is as follows: Figure 5a As shown, the node expansion of the Hybrid-A* algorithm in this embodiment of the invention is as follows: Figure 5b As shown.

[0096] In each search step, the position of the child node is determined by the vehicle's minimum turning radius R and the discrete number N of the front wheel steering angles. The larger N is, the more child node positions can be accessed, and the smoother the path. The method for defining the child node position is as follows: Figure 6 As shown. Therefore, the path cost function in this embodiment of the invention considers the path length cost, the cost of turning and reversing, and can adapt to different cleaning vehicles and cover various layouts of photovoltaic fields, thus expanding the applicable scenarios.

[0097] 3) Determine the evaluation function of the offline path search algorithm based on the heuristic cost function and the path cost function.

[0098] In summary, the offline path search algorithm Hybrid-A* in this embodiment of the invention searches in a grid map based on the evaluation function f(s) = g(s) + h(s).

[0099] S230. Perform offline path planning for the photovoltaic panel area based on the offline path search algorithm, target point coordinate location information, target point access status information, and target point cleaning status information.

[0100] In this embodiment of the invention, offline path planning is performed based on the offline path search algorithm determined above, and the status of the target point is updated according to the target point access status information and the target point cleaning status information, so as to determine whether the access ratio has been reached and end the path planning.

[0101] Specifically, offline path planning is performed on the photovoltaic panel area based on the offline path search algorithm, target point coordinate location information, target point access status information, and target point cleaning status information, including:

[0102] 1) Access the cleaning target points within the photovoltaic panel area according to the offline path search algorithm;

[0103] 2) When passing through a grid occupied by a cleaning operation target point, the target point access status information and target point cleaning status information of the grid are updated, and the access ratio of the current cleaning operation target point is calculated based on the target point access status information.

[0104] S240. Determine whether the access ratio of the current cleaning operation target point has reached the preset ratio threshold.

[0105] S250. If the access ratio of the current cleaning operation target point reaches the preset ratio threshold, the search will stop and the preliminary results of offline path planning will be obtained.

[0106] It should be understood that if the access ratio of the current cleaning operation target point does not reach the preset ratio threshold, the search will continue until the preset ratio threshold is reached.

[0107] Specifically, in this embodiment of the invention, when starting path search, a TargetPoint set and a VisitedPoint set are created. The initial number of TargetPoints is M (M is the number of grid cells occupied by the target point). Each time the path planned by Hybrid-A* passes through a grid cell occupied by a target point, the access flag of that grid is updated to "visited," and the target point heuristic value is cleared. Simultaneously, that grid cell is moved from the TargetPoint set to the VisitedPoint set. After this, the target point grid cell is completely identical to an empty grid cell except for the cleaning work flag, and can be freely passed through. Considering that it is impossible to complete the cleaning of photovoltaic panels with the same precision as the grid map in practice, when the target point grid access ratio reaches a customized P, the cleaning is considered complete, and a path from the current point to the task endpoint can be planned.

[0108] Therefore, in this embodiment of the invention, since the grid map of the photovoltaic power plant is known and the cleaning vehicle directly follows the path, the Hybrid-A* path planning stage can be performed offline. When a cleaning vehicle with different parameters is used or the photovoltaic panel layout changes, it is only necessary to reset the path planning algorithm parameters and perform offline planning again. This method in this embodiment of the invention ensures that the cleaning effect is not affected while adapting to changes in the vehicle and the actual environment.

[0109] Furthermore, compared to online path planning that relies on sensor-based positioning, the method of this invention has a smaller cumulative error, more stable lateral distance changes between the cleaning vehicle and the photovoltaic panel, and is less prone to cleaning blind spots. Existing sensor-based positioning technologies are prone to upper or lower blind spots, for example... Figure 7 As shown, the method of this embodiment of the invention has a smaller cumulative error and is less likely to produce cleaning blind spots because it sets the south boundary line and determines the target point at a distance K from the south boundary line.

[0110] In this embodiment of the invention, the preliminary results of the offline path planning are optimized according to the preset tracking rules of the photovoltaic unmanned cleaning vehicle, such as... Figure 8 As shown, it includes:

[0111] S310. Determine the preset tracking rules for the photovoltaic unmanned cleaning vehicle based on the curvature limit, the uniformity of curvature change, the uniformity of spacing, and the probability of collision.

[0112] S320. Identify the target points for abnormal cleaning operations based on the preset tracking rules in the offline path planning results.

[0113] S330. Optimize the offline path after removing abnormal cleaning operation target points to obtain the path planning results of the photovoltaic unmanned cleaning vehicle.

[0114] Specifically, the offline path after removing abnormal cleaning operation target points is optimized, including:

[0115] The offline path after removing abnormal cleaning target points is smoothed using the gradient descent method to increase smoothness.

[0116] The offline path with increased smoothness is densified using polynomial interpolation to obtain the path planning results for the photovoltaic unmanned cleaning vehicle.

[0117] It should be understood that, in the preliminary offline path planning results generated by the above algorithm, points with curvature exceeding the vehicle's minimum turning radius limit, uneven curvature changes, uneven spacing, and potential collisions are filtered out. Gradient descent is used to optimize the path and increase its smoothness. Considering that the unmanned cleaning vehicle does not have high requirements for ride comfort, cubic polynomial interpolation is used for densification processing to obtain a smooth, dense, and uniform path.

[0118] Finally, a real-vehicle test of the unmanned cleaning vehicle can be conducted to verify whether it can successfully track and complete the cleaning process according to the cleaning marks to meet the requirements for cleaning photovoltaic panels.

[0119] In summary, the path planning method for unmanned photovoltaic cleaning vehicles provided in this invention uses a two-dimensional grid map for offline path planning. By extracting and redefining grid information, the working target point of the unmanned cleaning vehicle is successfully located, avoiding manual annotation and topological path definition. This process can be effectively implemented using existing grid map processing technology. Furthermore, considering the characteristic that photovoltaic panels are both obstacles and cleaning targets, this invention employs a multi-target traversal offline path planning method to traverse the photovoltaic panels along the planned path, while avoiding large cleaning blind spots caused by the cumulative errors of online calculations by multiple modules. A heuristic function is defined using the actual distance between the current point and the target point and a custom heuristic value to achieve efficient path search. In addition, the path planning method for unmanned photovoltaic cleaning vehicles provided in this invention fully considers the influence of photovoltaic panel layout and vehicle parameters, improving the Hybrid-A* path planning algorithm. Only a few parameters need to be modified to adapt to different vehicle and photovoltaic field layouts.

[0120] As another embodiment of the present invention, a path planning device for a photovoltaic unmanned cleaning vehicle is provided, for implementing the path planning method for a photovoltaic unmanned cleaning vehicle described above, wherein, as Figure 9 As shown, the path planning device 10 for the photovoltaic unmanned cleaning vehicle includes:

[0121] The grid map analysis module 100 is used to analyze the preset grid map of the photovoltaic panel area to determine the target points for cleaning operations. Each target point for cleaning operations is configured with at least the target point coordinate location information, target point access status information, target point cleaning status information, and target point heuristic value information used to guide the cleaning sequence.

[0122] The multi-objective traversal path planning module 200 is used to determine the start and end points of the photovoltaic panel area according to the preset path planning task, and to perform offline path planning on the photovoltaic panel area based on the offline path search algorithm until the access ratio of the cleaning operation target point reaches the preset ratio threshold, thereby obtaining the preliminary results of the offline path planning. The evaluation function of the offline path search algorithm includes at least a heuristic function and a path cost function. The heuristic function is a cost function determined based on the target point heuristic value information of each cleaning operation target point, and the path cost function is a cost function determined based on the kinematic characteristics and path movement consumption of the photovoltaic unmanned cleaning vehicle.

[0123] The path post-processing module 300 is used to optimize the preliminary offline path planning results according to the preset tracking rules of the photovoltaic unmanned cleaning vehicle, so as to obtain the path planning results of the photovoltaic unmanned cleaning vehicle.

[0124] The path planning device for unmanned photovoltaic cleaning vehicles provided by this invention analyzes a preset grid map, traverses and extracts map information, and redefines the grid information of multiple target points. Based on the kinematic characteristics of the unmanned photovoltaic cleaning vehicle, path movement costs, and target point heuristics, a cost function is constructed. Then, offline path planning is performed offline multi-target traversal using the results of offline path search algorithm analysis. This achieves automatic path planning for multi-target traversal across the entire photovoltaic power plant scenario. This path planning device for unmanned photovoltaic cleaning vehicles is based on a preset grid map and operates offline, eliminating the need for real-time monitoring. Therefore, it can reduce accumulated errors and overcome cleaning blind spots, effectively improving the path planning capability of unmanned photovoltaic cleaning vehicles while simultaneously enhancing cleaning performance.

[0125] In this embodiment of the invention, the Dijkstra graph search algorithm is specifically used to traverse the grid map, obtain information on photovoltaic panels and obstacles, as well as map orientation. Based on the traversal results, the edges of the photovoltaic panels are determined and a series of target points are defined. Information is then added to the target points for subsequent path planning. Based on the map analysis results and cleaning task instructions, a multi-target point management strategy is adopted, combining vehicle kinematic parameters, costs, and heuristic information to perform Hybrid-A* offline path planning, taking into account both cleaning and obstacle avoidance. Based on the path planning results, points whose curvature, spacing, etc., do not meet the requirements for autonomous vehicle tracking are filtered out, and interpolation and gradient descent methods are used to optimize the path, generating path information that can be directly used for autonomous vehicle tracking.

[0126] In this embodiment of the invention, only the mature Dijkstra graph search algorithm is used to traverse a two-dimensional grid map for map information filtering and redefinition, eliminating the need for cumbersome manual topology map construction. Multi-target point setting is based on the geometric information of the photovoltaic panels, which is simple, readily available, and highly accurate. The input information is only a predetermined grid map, without relying on autonomous driving strategies involving multiple modules such as perception and positioning. It eliminates the need for high-performance computer hardware and high-cost sensors, significantly reducing errors and costs. Furthermore, traversing the photovoltaic panels using multiple target points and their preset information does not fix the cleaning order of the unmanned vehicle. Based on the Hybrid-A* method that considers vehicle parameters, it can adapt to different cleaning vehicles and cover various layouts of photovoltaic fields, expanding the applicable scenarios. Anomaly information filtering and optimization are performed on the initial path to ensure that the generated path can be directly used by the unmanned vehicle, simplifying the process and improving the effectiveness. The entire process is performed offline with a two-dimensional grid map as input, without requiring real-time performance. The logic is clear, with few steps, resulting in small accumulated errors and a relatively stable lateral distance between the vehicle and the photovoltaic panels, overcoming the cleaning blind spot problem.

[0127] As another embodiment of the present invention, a storage medium is provided, wherein computer instructions are stored, which, when loaded and executed by a processor, implement the path planning method for photovoltaic unmanned cleaning vehicles described above.

[0128] In this embodiment of the invention, a non-transitory computer-readable storage medium is provided. The computer-readable storage medium stores computer-executable instructions that can execute the path planning method for the photovoltaic unmanned cleaning vehicle in any of the above method embodiments. The storage medium may be a magnetic disk, optical disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk drive (HDD), or solid-state drive (SSD), etc.; the storage medium may also include combinations of the above types of memory.

[0129] As another embodiment of the present invention, an electronic device is provided, comprising a memory and a processor, the memory and the processor being communicatively connected, the memory being used to store computer instructions, and the processor being used to load and execute the computer instructions to implement the path planning method for photovoltaic unmanned cleaning vehicles described above.

[0130] like Figure 10As shown, the electronic device 90 may include: at least one processor 91, such as a CPU (Central Processing Unit), at least one communication interface 93, a memory 94, and at least one communication bus 92. The communication bus 92 is used to enable communication between these components. The communication interface 93 may include a display screen or a keyboard; optionally, the communication interface 93 may also include a standard wired interface or a wireless interface. The memory 94 may be high-speed RAM (Random Access Memory) or non-volatile memory, such as at least one disk storage device. Optionally, the memory 94 may also be at least one storage device located remotely from the aforementioned processor 91. The memory 94 stores application programs, and the processor 91 calls the program code stored in the memory 94 to execute any of the above-described method steps.

[0131] The communication bus 92 can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The communication bus 92 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 9 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0132] The memory 93 may include volatile memory, such as random-access memory (RAM); the memory may also include non-volatile memory, such as flash memory, hard disk drive (HDD) or solid-state drive (SSD); the memory 94 may also include a combination of the above types of memory.

[0133] Among them, processor 91 can be a central processing unit (CPU), a network processor (NP), or a combination of CPU and NP.

[0134] The processor 91 may further include a hardware chip. This hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The PLD may be a complex programmable logic device (CPLD), a field-programmable gate array (FPGA), a generic array logic (GAL), or any combination thereof.

[0135] Optionally, memory 94 is also used to store program instructions. Processor 91 can invoke program instructions to implement the present invention. Figure 1 The path planning method for photovoltaic unmanned cleaning vehicles is shown in the embodiments.

[0136] It is understood that the above embodiments are merely exemplary implementations used to illustrate the principles of the present invention, and the present invention is not limited thereto. For those skilled in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A path planning method for photovoltaic unmanned cleaning vehicles, characterized in that, include: The preset grid map of the photovoltaic panel area is analyzed to determine the target points for cleaning operations. Each target point for cleaning operations is configured with at least the target point coordinate location information, target point access status information, target point cleaning status information, and target point heuristic value information used to guide the cleaning sequence. The task start and end points of the photovoltaic panel area are determined according to the preset path planning task, and offline path planning is performed on the photovoltaic panel area based on the offline path search algorithm until the access ratio of the cleaning operation target point reaches the preset ratio threshold, and the preliminary results of offline path planning are obtained. The evaluation function of the offline path search algorithm includes at least a heuristic function and a path cost function. The heuristic function is a cost function determined based on the target point heuristic value information of each cleaning operation target point, and the path cost function is a cost function determined based on the kinematic characteristics and path movement consumption of the photovoltaic unmanned cleaning vehicle. The preliminary offline path planning results are optimized based on the preset tracking rules of the photovoltaic unmanned cleaning vehicle to obtain the path planning results of the photovoltaic unmanned cleaning vehicle. The analysis includes the pre-defined grid map of the photovoltaic panel area, including: The preset grid map is traversed and searched using a graph search algorithm to locate the positions of all photovoltaic panels in the photovoltaic panel area; The boundary line information of each photovoltaic panel is determined based on the position of each photovoltaic panel and the orientation information of the preset grid map. The boundary line information of each photovoltaic panel includes at least one boundary line along the length direction of the photovoltaic panel. Based on the boundary line information of each photovoltaic panel and the preset interval distance, multiple cleaning operation target points corresponding to each photovoltaic panel are determined; This involves determining multiple cleaning target points for each photovoltaic panel based on its boundary line information and preset intervals, including: The preset interval distance is determined based on the current dimensions of the photovoltaic unmanned cleaning vehicle and the preset cleaning range of the cleaning unit; The locations of multiple cleaning target points for each photovoltaic panel are determined based on the boundary line information of each photovoltaic panel and the preset interval distance; Among them, the location of each photovoltaic panel, combined with the map orientation, is the distance on the south side of the line. Define a target point with the same number of grid cells along the southern edge, at a distance of... The definition method is as follows: ， in, Indicates the number of grid cells. Defined based on the length, width, and height of the unmanned cleaning vehicle. The cleaning range of the unmanned cleaning vehicle is defined based on its cleaning capacity.

2. The path planning method for photovoltaic unmanned cleaning vehicles according to claim 1, characterized in that, The task start and end points of the photovoltaic panel area are determined according to the preset path planning task, and offline path planning is performed on the photovoltaic panel area based on the offline path search algorithm until the access ratio of the cleaning operation target point reaches a preset ratio threshold, and the preliminary results of offline path planning are obtained, including: The starting point for the unmanned photovoltaic cleaning vehicle to enter the photovoltaic panel area and the ending point for leaving the photovoltaic panel area are determined according to the preset path planning task. The evaluation function of the offline path search algorithm is determined based on the target point heuristic value of each cleaning operation target point, the kinematic characteristics of the photovoltaic unmanned cleaning vehicle, and the path movement consumption. Offline path planning is performed on the photovoltaic panel area based on the offline path search algorithm, target point coordinate location information, target point access status information, and target point cleaning status information. Determine whether the access ratio of the current cleaning operation target point has reached the preset ratio threshold; If the access ratio of the current cleaning operation target point reaches the preset ratio threshold, the search will stop and the preliminary results of offline path planning will be obtained.

3. The path planning method for photovoltaic unmanned cleaning vehicles according to claim 2, characterized in that, The evaluation function of the offline path search algorithm is determined based on the target point heuristic value of each cleaning operation target point, the kinematic characteristics of the photovoltaic unmanned cleaning vehicle, and the path movement consumption, including: The heuristic cost function is determined based on the target point heuristic value of each cleaning operation target point and the distance between the current grid point of the photovoltaic unmanned cleaning vehicle and the cleaning operation target point; The path length cost, turning cost, and reversing cost are determined based on the kinematic characteristics of the photovoltaic unmanned cleaning vehicle, and the path cost function is determined based on the path length cost, turning cost, and reversing cost. The evaluation function of the offline path search algorithm is determined based on the heuristic cost function and the path cost function.

4. The path planning method for photovoltaic unmanned cleaning vehicles according to claim 2, characterized in that, Based on the offline path search algorithm, target point coordinates, target point access status, and target point cleaning status, offline path planning is performed on the photovoltaic panel area, including: The cleaning target points within the photovoltaic panel area are accessed according to the offline path search algorithm. When passing through a grid occupied by a cleaning operation target point, the target point access status information and target point cleaning status information of that grid are updated, and the access ratio of the current cleaning operation target point is calculated based on the target point access status information.

5. The path planning method for photovoltaic unmanned cleaning vehicles according to claim 1, characterized in that, The preliminary results of the offline path planning are optimized according to the preset tracking rules of the photovoltaic unmanned cleaning vehicle, including: The preset tracking rules for the photovoltaic unmanned cleaning vehicle are determined based on the curvature limit, the uniformity of curvature change, the uniformity of spacing, and the probability of collision. Based on the preset tracking rules, the target points for abnormal cleaning operations are identified in the preliminary results of the offline path planning. The offline path after removing abnormal cleaning operation target points is optimized to obtain the path planning results of the photovoltaic unmanned cleaning vehicle.

6. The path planning method for photovoltaic unmanned cleaning vehicles according to claim 5, characterized in that, The offline path after removing abnormal cleaning operation target points is optimized, including: The offline path after removing abnormal cleaning target points is smoothed using the gradient descent method to increase smoothness. The offline path with increased smoothness is densified using polynomial interpolation to obtain the path planning results for the photovoltaic unmanned cleaning vehicle.

7. A path planning device for a photovoltaic unmanned cleaning vehicle, used to implement the path planning method for a photovoltaic unmanned cleaning vehicle as described in any one of claims 1 to 6, characterized in that, The path planning device for the photovoltaic unmanned cleaning vehicle includes: The grid map analysis module is used to analyze the preset grid map of the photovoltaic panel area to determine the target points for cleaning operations. Each target point for cleaning operations is configured with at least the target point coordinate location information, target point access status information, target point cleaning status information, and target point heuristic value information used to guide the cleaning sequence. The multi-objective traversal path planning module is used to determine the start and end points of the photovoltaic panel area according to the preset path planning task, and to perform offline path planning on the photovoltaic panel area based on the offline path search algorithm until the access ratio of the cleaning operation target point reaches the preset ratio threshold, and obtain the preliminary results of the offline path planning. The evaluation function of the offline path search algorithm includes at least a heuristic function and a path cost function. The heuristic function is a cost function determined based on the target point heuristic value information of each cleaning operation target point, and the path cost function is a cost function determined based on the kinematic characteristics and path movement consumption of the photovoltaic unmanned cleaning vehicle. The path post-processing module is used to optimize the preliminary offline path planning results according to the preset tracking rules of the photovoltaic unmanned cleaning vehicle, so as to obtain the path planning result of the photovoltaic unmanned cleaning vehicle.

8. A storage medium, characterized in that, Used to store computer instructions, which, when loaded and executed by a processor, implement the path planning method for a photovoltaic unmanned cleaning vehicle as described in any one of claims 1 to 6.

Images