Sweeper path planning method, system and equipment suitable for complex environment and medium

Abstract

The invention relates to a sweeper path planning method, system and device suitable for a complex environment and a medium. The method comprises the steps that a global map of a sweeping area is acquired, feature vertexes are determined to decompose the global map, and a structural sub-area and a special sub-area are generated; determining a starting point of the structure sub-region according to the feature vertex, and generating a first path covering the structure sub-region by adopting an internal spiral method based on the starting point; determining a non-linear constraint region of the special sub-region, converting the non-linear constraint region based on coordinate system transformation, and generating a second path of the special sub-region; and determining a starting point and a target point of any path, connecting the starting point and the target point based on an improved path search algorithm, and generating a target path of the cleaning area. Therefore, according to the invention, through accurate path planning and dynamic adaptability optimization, the path planning level and the completion quality of the sweeping operation are improved, so that the sweeping vehicle can adapt to the actual scene of a complex environment, and the application range is greatly expanded.

Description

Technical Field

[0001] This invention relates to the field of path planning technology, and specifically to a path planning method, system, device and medium for sweepers applicable to complex environments. Background Technology

[0002] With the development of autonomous driving technology, driverless sweeper technology has greatly reduced the pressure of urban sanitation and cleaning. At the same time, it has brought greater challenges to the full coverage path planning of sweepers, requiring the planned path to meet the requirements of low repetition coverage, short total length, few turning points and short time consumption.

[0003] However, existing research on full-coverage path planning mainly focuses on single-feature environments, with fewer studies on full-coverage path planning for complex environments with multiple features. This will significantly limit the practical application scenarios and cleaning effects of unmanned cleaning vehicles.

[0004] Furthermore, existing methods typically employ path planning based on raster maps, whose mapping quality depends on resolution and the shape of the environment, making them unsuitable for modeling complex environments. Moreover, the resulting paths often exhibit a multi-breakpoint polyline structure, which is unsuitable for sweepers with limited turning capabilities, leading to decreased operational stability. Clearly, existing technologies have shortcomings that urgently need to be addressed. Summary of the Invention

[0005] The primary objective of this application is to solve at least one of the above-mentioned problems by providing a sweeper path planning method and system applicable to complex environments.

[0006] To achieve the various objectives of this application, the following technical solution is adopted: A sweeper path planning method applicable to complex environments, provided to meet one of the purposes of this application, includes the following steps: Obtain a global map of the cleaning area, determine feature vertices to decompose the global map, and generate structural sub-regions and special sub-regions; The starting point of the structured sub-region is determined based on the feature vertices, and a first path covering the structured sub-region is generated based on the starting point using a covering algorithm. The nonlinear constraint region of a special sub-region is determined, and the nonlinear constraint region is transformed based on coordinate system transformation to generate a second path for the special sub-region; Determine the starting point and target point of any path, and connect the starting point and target point based on an improved path search algorithm to generate the target path of the cleaning area.

[0007] In an optional embodiment, in a first aspect of the invention, determining feature vertices to decompose the global map and generate structured sub-regions and special sub-regions includes the following steps: Parse the global map to obtain obstacle information and global map boundaries, determine local map boundaries based on obstacle information, and determine feature vertices based on either mandatory points or repeated points on the path in the global map. Preset region identification rules are used to determine the structural template of the decomposed region based on the global map boundary and the local map boundary. The structural template is then used to decompose the global map and generate structural sub-regions. Preset region decomposition constraints are applied to the global map of the structure template decomposition based on minimizing the distance to feature vertices of all decomposed regions. Unstructured sub-regions are configured as special sub-regions based on global map boundaries and path characteristics.

[0008] In an optional embodiment, in a first aspect of the invention, determining the starting point of a structural sub-region based on feature vertices includes the following steps: Based on a preset coverage algorithm, several starting points that satisfy the motion rules corresponding to the coverage algorithm are generated in any structural sub-region. A preset starting point distance constraint is used to determine the starting point of all structural sub-regions based on minimizing the distance between the starting points of adjacent structural sub-regions.

[0009] In an optional embodiment, in a first aspect of the invention, generating a first path for a covered sub-region based on a starting point using a covering algorithm includes the following steps: Construct a grid map of all structural sub-regions, determine motion rules according to a preset coverage algorithm, and for each structural sub-region, traverse the grid map from the starting point according to the motion rules to generate a coverage path, wherein the coverage path includes a starting point and an ending point; A predetermined endpoint distance constraint is set between adjacent structural sub-regions. The target point in the coverage path is determined based on the minimum distance between the endpoint of the coverage path and the starting point of the adjacent structural sub-region. The first path is determined based on the starting point, the target point, and the corresponding coverage path.

[0010] In an optional embodiment, in a first aspect of the invention, determining the nonlinear constraint region of a specific sub-region includes the following steps: Point cloud data of a specific sub-region is acquired, the point cloud data is converted to a unified coordinate system, and roadside geometric parameters are extracted in the unified coordinate system to determine the nonlinear constraint region.

[0011] In an optional embodiment, in a first aspect of the invention, the nonlinear constraint region is transformed based on coordinate system transformation to generate a second path for a special sub-region, comprising the following steps: For the nonlinear constraint region, key points are modeled by transforming the unified coordinate system into the Frenet coordinate system; The number of key points is determined based on the maximum width of the nonlinear constraint region and the width of the sweeper. The key points are then updated through a quadratic coordinate transformation based on the positional order of the key points within the nonlinear constraint region to determine the target key points in a unified coordinate system. A Bézier curve is introduced to smooth the target key points under the nonlinear constraint region to generate a second path for a special sub-region. The second path includes a starting point and a target point. The expression for the key points in Frenet coordinates is: ; ; In the formula, This represents the x-coordinate of the i-th key point. This represents the ordinate of the i-th key point. This represents the coordinates of the left vertex of the nonlinear constraint region. This represents the coordinates of the right vertex of the nonlinear constraint region, where N represents the number of key points. Represents key points in Frenet coordinates. Represents key points in a unified coordinate system; The expression for smoothing the Bézier curve is: ; In the formula, , , , The continuous inflection points represent the control points, and t represents the proportional coefficient.

[0012] In an optional embodiment, in a first aspect of the invention, determining the starting point and target point of an arbitrary path, and connecting the starting point and target point based on an improved path search algorithm to generate a target path includes the following steps: Based on the Hybrid A* algorithm, an evaluation function is set to minimize the cost of connecting the target point to the starting point of adjacent paths. The target path is generated by connecting the starting point and the target point in all first and second paths. The evaluation function of the Hybrid A* algorithm is expressed as follows: ; ; In the formula, This represents the evaluation function of the Hybrid A* algorithm. This represents the actual cumulative cost from the target point to the current starting point n. This represents the heuristic value based on Euclidean distance. This represents the cost of calculating distance using the Reeds-Shepp curve, with max indicating the value of taking... or The maximum value in the heuristic function is used as the cost. The actual cost represents the actual path length from the target point to the current starting point. This indicates the cost of the shift or change. This indicates the cost of changing gears. This indicates the cost of reversing. This indicates the weight of the actual cost. The weights representing the shifts in direction. This indicates the weight of gear shifting. Indicates the weight of reversing. The weight represents the actual cumulative cost of the i-th current starting point n. This represents the actual cumulative cost of the i-th current starting point n.

[0013] On the other hand, a sweeper path planning system suitable for complex environments, provided to meet one of the purposes of this application, includes: The global decomposition module is used to obtain a global map of the cleaning area, determine feature vertices to decompose the global map, and generate structural sub-regions and special sub-regions. The first planning module is used to determine the starting point of the structured sub-region based on the feature vertices, and to generate the first path covering the structured sub-region based on the starting point using a covering algorithm. The second planning module is used to determine the nonlinear constraint region of a special sub-region, transform the nonlinear constraint region based on coordinate system transformation, and generate the second path of the special sub-region. The path generation module is used to determine the starting point and target point of any path, and connect the starting point and target point based on an improved path search algorithm to generate the target path of the cleaning area.

[0014] In an optional embodiment, in a second aspect of the invention, determining feature vertices to decompose the global map and generate structural sub-regions and special sub-regions includes the following steps: Parse the global map to obtain obstacle information and global map boundaries, determine local map boundaries based on obstacle information, and determine feature vertices based on either mandatory points or repeated points on the path in the global map. Preset region identification rules are used to determine the structural template of the decomposed region based on the global map boundary and the local map boundary. The structural template is then used to decompose the global map and generate structural sub-regions. Preset region decomposition constraints are applied to the global map of the structure template decomposition based on minimizing the distance to feature vertices of all decomposed regions. Unstructured sub-regions are configured as special sub-regions based on global map boundaries and path characteristics.

[0015] In an optional embodiment, in a second aspect of the invention, determining the starting point of the structural sub-region based on the feature vertices includes the following steps: Based on a preset coverage algorithm, several starting points that satisfy the motion rules corresponding to the coverage algorithm are generated in any structural sub-region. A preset starting point distance constraint is used to determine the starting point of all structural sub-regions based on minimizing the distance between the starting points of adjacent structural sub-regions.

[0016] In an optional embodiment, in a second aspect of the invention, generating a first path for a covered sub-region based on a starting point using a covering algorithm includes the following steps: Construct a grid map of all structural sub-regions, determine motion rules according to a preset coverage algorithm, and for each structural sub-region, traverse the grid map from the starting point according to the motion rules to generate a coverage path, wherein the coverage path includes a starting point and an ending point; A predetermined endpoint distance constraint is set between adjacent structural sub-regions. The target point in the coverage path is determined based on the minimum distance between the endpoint of the coverage path and the starting point of the adjacent structural sub-region. The first path is determined based on the starting point, the target point, and the corresponding coverage path.

[0017] In an optional embodiment, in a second aspect of the invention, determining the nonlinear constraint region of a specific sub-region includes the following steps: Point cloud data of a specific sub-region is acquired, the point cloud data is converted to a unified coordinate system, and roadside geometric parameters are extracted in the unified coordinate system to determine the nonlinear constraint region.

[0018] In an optional embodiment, in a second aspect of the invention, the nonlinear constraint region is transformed based on coordinate system transformation to generate a second path for a special sub-region, comprising the following steps: For the nonlinear constraint region, key points are modeled by transforming the unified coordinate system into the Frenet coordinate system; The number of key points is determined based on the maximum width of the nonlinear constraint region and the width of the sweeper. The key points are then updated through a quadratic coordinate transformation based on the positional order of the key points within the nonlinear constraint region to determine the target key points in a unified coordinate system. A Bézier curve is introduced to smooth the target key points under the nonlinear constraint region to generate a second path for a special sub-region. The second path includes a starting point and a target point. The expression for the key points in Frenet coordinates is: ; ; In the formula, This represents the x-coordinate of the i-th key point. This represents the ordinate of the i-th key point. This represents the coordinates of the left vertex of the nonlinear constraint region. This represents the coordinates of the right vertex of the nonlinear constraint region, where N represents the number of key points. Represents key points in Frenet coordinates. Represents key points in a unified coordinate system; The expression for smoothing the Bézier curve is: ; In the formula, , , , The continuous inflection points represent the control points, and t represents the proportional coefficient.

[0019] In an optional embodiment, in a second aspect of the invention, determining the starting point and target point of an arbitrary path, and connecting the starting point and target point based on an improved path search algorithm to generate a target path includes the following steps: Based on the Hybrid A* algorithm, an evaluation function is set to minimize the cost of connecting the target point to the starting point of adjacent paths. The target path is generated by connecting the starting point and the target point in all first and second paths. The evaluation function of the Hybrid A* algorithm is expressed as follows: ; ; In the formula, This represents the evaluation function of the Hybrid A* algorithm. This represents the actual cumulative cost from the target point to the current starting point n. This represents the heuristic value based on Euclidean distance. This represents the cost of calculating distance using the Reeds-Shepp curve, with max indicating the value of taking... or The maximum value in the heuristic function is used as the cost. The actual cost represents the actual path length from the target point to the current starting point. This indicates the cost of the shift or change. This indicates the cost of changing gears. This indicates the cost of reversing. This indicates the weight of the actual cost. The weights representing the shifts in direction. This indicates the weight of gear shifting. Indicates the weight of reversing. The weight represents the actual cumulative cost of the i-th current starting point n. This represents the actual cumulative cost of the i-th current starting point n.

[0020] On another front, a sweeper path planning device suitable for complex environments, provided to meet one of the purposes of this application, includes a central processing unit and a memory, wherein the central processing unit is used to invoke and run a computer program stored in the memory to execute the steps of the sweeper path planning method suitable for complex environments described in this application.

[0021] In another aspect, a computer-readable storage medium is provided to suit one of the purposes of this application, the computer-readable storage medium storing computer-executable instructions for causing a computer to perform a sweeper path planning method applicable to complex environments as disclosed in any of the first aspects of the present invention.

[0022] The technical solution of this application has many advantages, including but not limited to the following aspects: This application decomposes complex cleaning areas into independently processable modules through feature vertex extraction and sub-region division. For structured sub-regions, a coverage algorithm generates the first path, ensuring full coverage of regular areas and minimizing path redundancy. For special sub-regions, coordinate system transformation of nonlinear constraint areas converts complex constraints into linear problems, generating a more adaptable second path. An improved path search algorithm generates a globally optimal path by connecting arbitrary starting and ending points, reducing path fragmentation. The algorithm dynamically adjusts search weights based on path features from structured and special sub-regions, ensuring path smoothness and feasibility in complex scenarios. The hierarchical planning strategy not only reduces computational complexity and shortens path planning time, adapting to real-time adjustments in dynamic environments, but also balances cleaning efficiency and path safety, reducing the risk of path interruption due to environmental changes. Especially for sweepers with weak turning performance, precise path planning and dynamic adaptive optimization significantly reduce manual intervention costs, improve the automation level and completion quality of cleaning operations, enabling sweepers to adapt to complex environmental application scenarios and greatly expanding their application scope. Attached Figure Description

[0023] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is an exemplary architectural diagram of this application; Figure 2A flowchart illustrating one embodiment of the sweeper path planning method applicable to complex environments according to this application; Figure 3 This is a schematic diagram of global map decomposition in the sweeper path planning method applicable to complex environments adopted in this application; Figure 4 This is a schematic diagram illustrating the key point modeling in the sweeper path planning method applicable to complex environments used in this application. Detailed Implementation

[0024] 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 of the present invention. 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 are within the scope of protection of the present invention.

[0025] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0026] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0027] This invention discloses a path planning method, system, device, and medium for sweepers applicable to complex environments. The method decomposes complex cleaning areas into independent processing modules through feature vertex extraction and sub-region division. For structural sub-regions, a coverage algorithm generates a first path, ensuring full coverage of regular areas and minimizing path redundancy. For special sub-regions, coordinate system transformation of nonlinear constraint areas converts complex constraints into linear problems, generating a more adaptable second path. An improved path search algorithm generates a globally optimal path by connecting arbitrary starting and ending points, reducing path fragmentation. The algorithm dynamically adjusts search weights based on path features from structural and special sub-regions, ensuring path smoothness and feasibility in complex scenarios. The hierarchical planning strategy not only reduces computational complexity and shortens path planning time, adapting to real-time adjustments in dynamic environments, but also balances cleaning efficiency and path safety, reducing the risk of path interruption due to environmental changes. Especially for sweepers with weak turning performance, precise path planning and dynamic adaptive optimization significantly reduce manual intervention costs, improve the automation level and completion quality of cleaning operations, enabling sweepers to adapt to complex environmental application scenarios and greatly expanding their application scope. The following sections will provide detailed explanations.

[0028] The specific embodiments described below can be combined with each other, and the same or similar concepts or processes will not be described again in some embodiments. The embodiments of this application will be described below with reference to the accompanying drawings.

[0029] See Figure 1 This application discloses a path planning method for a sweeper vehicle applicable to complex environments. In its typical stress test embodiment, it includes the following steps: 101. Obtain a global map of the cleaning area, determine feature vertices to decompose the global map, and generate structural sub-regions and special sub-regions; It should be noted that the complex environment proposed in this application includes several nonlinear constraint regions and obstacles within these regions. Typically, path planning for the cleaning area first involves collecting map information to construct a map of the known environment. This can be done using electronic maps, SALM algorithms, coordinate acquisition devices, or laser scanners. Then, the acquired map information is used to create a gridded map, and obstacles are placed within the gridded map. Next, an appropriate full-coverage path planning algorithm (such as the inner spiral coverage method, reciprocating coverage method, or random coverage method) is selected to traverse the gridded map to search for nearby points of the uncleaned areas, and obstacle avoidance operations are performed based on the pre-set obstacles, thereby realizing path planning for the area to be cleaned.

[0030] Furthermore, such as Figure 3As shown, existing methods typically employ path planning based on grid maps, whose mapping effectiveness depends on resolution and the shape of the environment, making them unsuitable for modeling complex environments. Furthermore, the resulting paths often exhibit multi-breakpoint polyline structures, which are unsuitable for sweepers with limited turning capabilities, leading to decreased operational stability. Therefore, it is necessary to decompose the global map to plan sweeper paths within nonlinear constraint areas locally. Understandably, to improve path planning efficiency, generating straight paths with fewer inflection points is suitable for most sweeper types, especially for large sweepers with limited turning capabilities, where the efficiency improvement in path planning is even more significant.

[0031] Optionally, the feature vertex can be a garbage collection station or garbage treatment station in the cleaning area used for garbage collection or dumping. The sweeper needs to use the feature vertex as the starting point and return point for garbage handling. In addition to garbage stations, the feature vertex can also be other locations that cooperate with the cleaning task and have a round-trip nature. The specific configuration can be made by those skilled in the art according to the actual cleaning situation.

[0032] 102. Determine the starting point of the structured sub-region based on the feature vertices, and generate the first path covering the structured sub-region based on the starting point using a covering algorithm.

[0033] Optionally, the covering algorithm can be one of the following: inner spiral covering method, reciprocating covering method, or random covering method. The inner spiral covering method uses a spiral motion rule to traverse the grid map points in the region to generate a planned path, while the reciprocating covering method uses a bow-shaped or zigzag motion rule to traverse the grid map points in the region to generate a planned path. Those skilled in the art can select the corresponding covering algorithm according to the characteristics of the structural sub-region and the required covering method.

[0034] 103. Determine the nonlinear constraint region of the special sub-region, transform the nonlinear constraint region based on coordinate system transformation, and generate the second path of the special sub-region.

[0035] It should be noted that the special sub-regions are unstructured areas that require comprehensive consideration of C-shaped paths, S-shaped paths, sidewalks, garbage stations, and other irregular obstacles. Correspondingly, the nonlinear constraint areas are special areas where path planning cannot be solved by generating planned paths through straight-line motion trajectory constraints and requires nonlinear constraints (such as S-shaped paths).

[0036] 104. Determine the starting point and target point of any path, and connect the starting point and target point based on the improved path search algorithm to generate the target path of the cleaning area.

[0037] As can be seen, through the above embodiments, the complex cleaning area is decomposed into independent processing modules by feature vertex extraction and sub-region division. The structured sub-region uses a coverage algorithm to generate the first path, ensuring full coverage of the regular area and minimizing path redundancy. The special sub-region uses coordinate system transformation of the nonlinear constraint area to transform the complex constraints into a linear problem, generating a more adaptable second path. The improved path search algorithm generates a globally optimal path by connecting any starting point and the target point, reducing path fragmentation. The algorithm combines the path features of the structured and special sub-regions and dynamically adjusts the search weights to ensure the smoothness and feasibility of the path in complex scenarios. Among them, the hierarchical planning strategy not only reduces computational complexity and shortens path planning time, adapting to the real-time adjustment needs of dynamic environments, but also balances cleaning efficiency and path safety, reducing the risk of path interruption due to environmental changes. Especially for sweepers with weak turning performance, precise path planning and dynamic adaptive optimization significantly reduce the cost of manual intervention, improve the automation level and completion quality of cleaning operations, and enable sweepers to adapt to real-world application scenarios in complex environments, greatly expanding the application scope.

[0038] As an optional embodiment, the step of determining feature vertices to decompose the global map and generate structural sub-regions and special sub-regions in the above steps includes the following steps: Parse the global map to obtain obstacle information and global map boundaries, determine local map boundaries based on obstacle information, and determine feature vertices based on either mandatory points or repeated points on the path in the global map. Preset region identification rules are used to determine the structural template of the decomposed region based on the global map boundary and the local map boundary. The structural template is then used to decompose the global map and generate structural sub-regions. Preset region decomposition constraints are applied to the global map of the structure template decomposition based on minimizing the distance to feature vertices of all decomposed regions. Unstructured sub-regions are configured as special sub-regions based on global map boundaries and path characteristics.

[0039] Specifically, the structural template is to globally Figure 2 The standard shape template for dimensionalization, where the structure template is represented as a regular shape such as a rectangle on a two-dimensional surface. The structure template can also be represented as a structure region type that can complete path planning only through linear constraints. It can be used to repeatedly generate several structure sub-regions to improve the efficiency of path planning.

[0040] As can be seen, through the above embodiments, by parsing the obstacle information and boundaries of the global map and combining necessary points or path repetition points as feature vertices, key nodes in complex environments can be identified. The selection of feature vertices is based on path characteristics, ensuring that the subsequent decomposed structural sub-regions are strongly correlated with actual cleaning needs. Preset region identification rules are used to construct structural templates based on global and local boundaries, dividing the global map into easily covered structural sub-regions. Preset decomposition constraints ensure that the division of structural sub-regions meets the requirements of path continuity and coverage efficiency, reducing redundant paths or coverage blind spots caused by blind segmentation. The combined application of feature vertices and structural templates makes map decomposition conform to both physical boundaries and path characteristics, reducing the coverage loss problem caused by rule template mismatch in traditional methods. By minimizing the distance to feature vertices, the compactness of structural sub-region division and path continuity are ensured, reducing the energy consumption of frequent turning or repeated cleaning by the sweeper. Furthermore, the dynamic configuration mechanism of special sub-regions effectively solves the path planning problem of irregular obstacles or complex environments, improving the passability of sweepers in complex scenarios.

[0041] As an optional embodiment, the step of determining the starting point of the structural sub-region based on the feature vertices in the above steps includes the following steps: Based on a preset coverage algorithm, several starting points that satisfy the motion rules corresponding to the coverage algorithm are generated in any structural sub-region. A preset starting point distance constraint is used to determine the starting point of all structural sub-regions based on minimizing the distance between the starting points of adjacent structural sub-regions.

[0042] Optionally, the covering algorithm can be one of the following: inner spiral covering method, reciprocating covering method, or random covering method. The inner spiral covering method uses a spiral motion rule to traverse the grid map points in the region to generate a planned path, while the reciprocating covering method uses a bow-shaped or zigzag motion rule to traverse the grid map points in the region to generate a planned path. Those skilled in the art can select the corresponding covering algorithm according to the characteristics of the structural sub-region and the required covering method.

[0043] Specifically, the global map is decomposed into several structural sub-regions. For any structural sub-region, in order to meet the problem of short total length and low repetition coverage of the sweeper's full coverage path planning, it is necessary to consider the starting point of the first path in the connected structural sub-regions, thereby shortening the actual total length of the first path and improving the efficiency and feasibility of path planning.

[0044] As can be seen from the above embodiments, a number of candidate starting points satisfying the motion rules within a sub-region are generated through a preset coverage algorithm, ensuring that the cleaning path conforms to the geometric constraints of the algorithm itself. The generation of candidate starting points is directly related to the motion mode of the coverage algorithm, avoiding path breakage or repeated coverage due to improper selection of starting points. Smooth connection of the global path is achieved through preset starting point distance constraints. The selection of starting points simultaneously satisfies the local optimization requirements of the coverage algorithm and the coherence requirements of the global path, forming a multi-objective collaborative optimization mechanism. Through the joint optimization of motion rules and distance constraints, the feasibility of the cleaning path in complex scenarios is ensured.

[0045] As an optional embodiment, the step above, generating the first path of the covered structure sub-region based on the starting point using a covering algorithm, includes the following steps: Construct a grid map of all structural sub-regions, determine motion rules according to a preset coverage algorithm, and for each structural sub-region, traverse the grid map from the starting point according to the motion rules to generate a coverage path, wherein the coverage path includes a starting point and an ending point; A predetermined endpoint distance constraint is set between adjacent structural sub-regions. The target point in the coverage path is determined based on the minimum distance between the endpoint of the coverage path and the starting point of the adjacent structural sub-region. The first path is determined based on the starting point, the target point, and the corresponding coverage path.

[0046] Specifically, for any first path, it is necessary to consider the full coverage of the first path between adjacent structural sub-regions. Therefore, it is necessary to consider the problem of shortening the total path length in the first path connection process. Among the generated several feasible coverage paths, the optimal endpoint is determined as the target point based on the position of the endpoint and the starting point to form the optimal target path. It can be understood that the first paths between adjacent structural sub-regions are interconnected. The selection of the first path in the first structural sub-region will affect the selection of the first path in the adjacent second structural sub-region.

[0047] As can be seen, through the above embodiments, by constructing a grid map of all structural sub-regions, the continuous space is discretized into computable grid cells, ensuring a high degree of matching between the coverage path and the physical environment, and avoiding coverage blind spots or path conflicts caused by spatial abstraction errors. A preset endpoint distance constraint is implemented, minimizing the physical distance between the endpoint of the coverage path in adjacent structural sub-regions and the starting point of the next region, achieving a smooth transition between regions. The coverage path generation simultaneously satisfies the coverage integrity within sub-regions and the minimum movement distance for inter-regional connections, forming a global-local collaborative optimization mechanism. Through the joint configuration of the starting point, target point, and coverage path, the sweeper's path throughout the cleaning area is ensured to be compact and efficient, avoiding the global path fragmentation problem caused by local optimal selection in traditional methods. This significantly improves the coverage and efficiency of cleaning operations, reduces energy consumption, and enhances the sweeper's path planning adaptability to complex environments.

[0048] As an optional embodiment, the step of determining the nonlinear constraint region of the special sub-region in the above steps includes the following steps: Point cloud data of a specific sub-region is acquired, the point cloud data is converted to a unified coordinate system, and roadside geometric parameters are extracted in the unified coordinate system to determine the nonlinear constraint region.

[0049] Optionally, the unified coordinate system is the Cartesian coordinate system. The point cloud data is transformed into the unified coordinate system to eliminate coordinate deviations between multiple source sensors or heterogeneous data, ensuring the accuracy and consistency of subsequent geometric parameter extraction.

[0050] Specifically, by acquiring point cloud data of special sub-regions, three-dimensional spatial information such as obstacles and terrain undulations is captured, providing high-resolution basic data for modeling nonlinear constraint areas. Geometric parameters of the roadside (such as edges, curvature, and slope) are extracted under a unified coordinate system, and the nonlinear characteristics of special sub-regions (such as irregular obstacle distribution) are quantified. The extraction of geometric parameters makes the boundary definition of nonlinear constraint areas more accurate, avoiding the path planning failure problem caused by ignoring terrain details in traditional methods. The determination of nonlinear constraint areas is based on the dynamic analysis of roadside geometric parameters (such as curvature threshold judgment and edge continuity detection), enabling the sweeper to identify complex terrain in real time. Furthermore, the accurate identification of nonlinear constraint areas provides a reliable obstacle avoidance basis for subsequent path search algorithms, reducing path conflicts or repeated planning problems caused by misjudgment of constraint areas.

[0051] As can be seen, through the above embodiments, the combination of point cloud data and a unified coordinate system significantly improves the geometric description accuracy of nonlinear constraint regions and enhances the modeling accuracy of nonlinear constraint regions in special sub-regions. The extraction and analysis of roadside geometric parameters enable the system to flexibly respond to irregular obstacles or dynamic environmental changes, improving dynamic adaptability. The accurate identification of nonlinear constraint regions effectively improves the working stability of sweepers in complex terrains. The accurate modeling of nonlinear constraint regions significantly reduces the path planning failure rate, improves sweeping efficiency and equipment safety, and is especially suitable for sweeper operations in complex environments. This further improves the automation level and completion quality of sweeping operations, enabling sweepers to adapt to actual application scenarios in complex environments and greatly expanding the application scope.

[0052] As an optional embodiment, such as Figure 4 As shown, the step above, transforming the nonlinear constraint region based on coordinate system transformation to generate a second path for a special sub-region, includes the following steps: For the nonlinear constraint region, key points are modeled by transforming the unified coordinate system into the Frenet coordinate system; The number of key points is determined based on the maximum width of the nonlinear constraint region and the width of the sweeper. The key points are then updated through a quadratic coordinate transformation based on the positional order of the key points within the nonlinear constraint region to determine the target key points in a unified coordinate system. A Bézier curve is introduced to smooth the target key points under the nonlinear constraint region to generate a second path for a special sub-region. The second path includes a starting point and a target point. The expression for the key points in Frenet coordinates is: ; ; In the formula, This represents the x-coordinate of the i-th key point. This represents the ordinate of the i-th key point. This represents the coordinates of the left vertex of the nonlinear constraint region. This represents the coordinates of the right vertex of the nonlinear constraint region, where N represents the number of key points. Represents key points in Frenet coordinates. Represents key points in a unified coordinate system; The expression for smoothing the Bézier curve is: ; In the formula, , , , The continuous inflection points represent the control points, and t represents the proportional coefficient.

[0053] As can be seen, through the above embodiments, by transforming the nonlinear constraint region from a unified coordinate system to a Frenet coordinate system, a dynamic parameterized description of complex terrain is achieved. The introduction of the Frenet coordinate system enables key point modeling to be directly related to the curvature characteristics of the path, avoiding the path distortion problem caused by coordinate offset in the traditional Cartesian coordinate system. It can also transform the nonlinear constraint problem under the original unified coordinate system into a linear constraint problem for key point solving. The number of key points is dynamically determined according to the maximum width of the nonlinear constraint region and the width of the sweeper, ensuring that the path coverage matches the physical size of the equipment. The key point positions are updated through a quadratic coordinate transformation to ensure that the key points strictly follow the geometric boundaries of the nonlinear constraint region, avoiding path deviation or conflict. Bézier curves are introduced to the target key points for smoothing, generating a continuous and differentiable second path, significantly reducing the sweeper's sharp turns or frequent steering in complex terrain, allowing the path to flexibly adapt to the shape of the nonlinear constraint region (such as S-shaped bends and U-shaped obstacles), while preserving the connectivity between the starting point and the target point, ensuring the global feasibility of the path. This enables path planning to adapt to nonlinear constraint regions of different shapes, improving the stability and efficiency of the sweeper's operation. Consequently, the sweeper can adapt to real-world application scenarios in complex environments, significantly expanding its application scope.

[0054] As an optional embodiment, the steps described above, including determining the starting point and target point of any path, and connecting the starting point and target point based on an improved path search algorithm to generate the target path, include the following steps: Based on the Hybrid A* algorithm, an evaluation function is set to minimize the cost of connecting the target point to the starting point of adjacent paths. The target path is generated by connecting the starting point and the target point in all first and second paths. The evaluation function of the Hybrid A* algorithm is expressed as follows: ; ; In the formula, This represents the evaluation function of the Hybrid A* algorithm. This represents the actual cumulative cost from the target point to the current starting point n. This represents the heuristic value based on Euclidean distance. This represents the cost of calculating distance using the Reeds-Shepp curve, with max indicating the value of taking... or The maximum value in the heuristic function is used as the cost. The actual cost represents the actual path length from the target point to the current starting point. This indicates the cost of the shift or change. This indicates the cost of changing gears. This indicates the cost of reversing. This indicates the weight of the actual cost. The weights representing the shifts in direction. This indicates the weight of gear shifting. Indicates the weight of reversing. The weight represents the actual cumulative cost of the i-th current starting point n. This represents the actual cumulative cost of the i-th current starting point n.

[0055] Optionally, the visual sensors installed on the unmanned sweeper can be front-facing cameras, side-facing cameras, rear-facing cameras, surround-view cameras, or infrared cameras, depending on the actual scenario of the sweeping task and the sweeper's movement. For example, when performing nighttime sweeping tasks, infrared cameras can be used to acquire obstacle images, or when using a fishtail turn that requires reversing, front / rear-facing cameras or surround-view cameras can be used to acquire obstacle images. This application does not limit the type of visual sensor for the sweeper.

[0056] Optionally, the radar sensor installed on the unmanned sweeper can be a millimeter-wave radar, a two-dimensional lidar, a three-dimensional lidar, a solid-state lidar, or a mechanical lidar. Those skilled in the art can adjust the selection of radar sensor type and installation location according to the sweeping task, and this application does not impose any limitations.

[0057] Specifically, the Hybrid A* algorithm's evaluation function comprehensively considers multiple dimensions of costs, including path length, steering changes, gear shifts, and reversing, avoiding the suboptimal path problem caused by the single distance cost in the traditional A* algorithm. The actual weighted combination allows the algorithm to adapt to different scenario requirements. The path length cost is used to ensure the overall path is the shortest and reduce invalid movement. The steering change cost is used to reduce the impact of frequent steering on mechanical wear and improve operational stability. The gear shift cost is used to reduce gear shifting operations and optimize power system efficiency. The reversing cost is used to avoid frequent reversing and reduce energy consumption and path complexity. The flexible configuration of weight parameters allows the algorithm to dynamically balance different task objectives (such as prioritizing cleaning efficiency or equipment protection).

[0058] Furthermore, the heuristic function combines Euclidean distance and Reeds-Shepp curve distance to enhance robustness. Euclidean distance provides fast global direction guidance to ensure search efficiency, while Reeds-Shepp curve distance calculates the feasible path length based on vehicle kinematics models (such as minimum turning radius), avoiding the path infeasibility problem caused by neglecting steering constraints in traditional Euclidean distance. By selecting the values, the heuristic function is ensured to always provide a conservative but reliable lower bound estimate, avoiding premature pruning and missing the globally optimal path.

[0059] As can be seen, through the above embodiments, the improved Hybrid A* algorithm and multi-dimensional cost evaluation function design achieve the coordinated optimization of global optimality and local smoothness of path planning in complex environments, significantly improving the path continuity and movement feasibility of the sweeper when switching between structural sub-regions and special sub-regions. By connecting the starting point and target point of the first path (structural sub-region coverage path) and the second path (special sub-region obstacle avoidance path), a global target path is generated, ensuring seamless transition of the sweeper between regular and complex areas. This significantly reduces ineffective movement and equipment wear in path planning, improves sweeping efficiency and safety in complex environments, and is especially suitable for complex environmental operation scenarios that require frequent switching of movement modes.

[0060] Please see Figure 2 According to one aspect of this application, a sweeper path planning system suitable for complex environments is provided, the system comprising: The global decomposition module is used to obtain a global map of the cleaning area, determine feature vertices to decompose the global map, and generate structural sub-regions and special sub-regions. The first planning module is used to determine the starting point of the structured sub-region based on the feature vertices, and to generate the first path covering the structured sub-region based on the starting point using a covering algorithm. The second planning module is used to determine the nonlinear constraint region of a special sub-region, transform the nonlinear constraint region based on coordinate system transformation, and generate the second path of the special sub-region. The path generation module is used to determine the starting point and target point of any path, and connect the starting point and target point based on an improved path search algorithm to generate the target path of the cleaning area.

[0061] As can be seen, through the above embodiments, the complex cleaning area is decomposed into independent processing modules by feature vertex extraction and sub-region division. The structured sub-region uses a coverage algorithm to generate the first path, ensuring full coverage of the regular area and minimizing path redundancy. The special sub-region uses coordinate system transformation of the nonlinear constraint area to transform the complex constraints into a linear problem, generating a more adaptable second path. The improved path search algorithm generates a globally optimal path by connecting any starting point and the target point, reducing path fragmentation. The algorithm combines the path features of the structured and special sub-regions and dynamically adjusts the search weights to ensure the smoothness and feasibility of the path in complex scenarios. Among them, the hierarchical planning strategy not only reduces computational complexity and shortens path planning time, adapting to the real-time adjustment needs of dynamic environments, but also balances cleaning efficiency and path safety, reducing the risk of path interruption due to environmental changes. Especially for sweepers with weak turning performance, precise path planning and dynamic adaptive optimization significantly reduce the cost of manual intervention, improve the automation level and completion quality of cleaning operations, and enable sweepers to adapt to real-world application scenarios in complex environments, greatly expanding the application scope.

[0062] Based on any embodiment of the system in this application, as an optional embodiment, determining feature vertices to decompose the global map and generate structural sub-regions and special sub-regions includes the following steps: Parse the global map to obtain obstacle information and global map boundaries, determine local map boundaries based on obstacle information, and determine feature vertices based on either mandatory points or repeated points on the path in the global map. Preset region identification rules are used to determine the structural template of the decomposed region based on the global map boundary and the local map boundary. The structural template is then used to decompose the global map and generate structural sub-regions. Preset region decomposition constraints are applied to the global map of the structure template decomposition based on minimizing the distance to feature vertices of all decomposed regions. Unstructured sub-regions are configured as special sub-regions based on global map boundaries and path characteristics.

[0063] As can be seen, through the above optional embodiments, by parsing the obstacle information and boundaries of the global map and combining necessary points or path repetition points as feature vertices, key nodes in complex environments can be identified. The selection of feature vertices is based on path characteristics, ensuring that the subsequent decomposed structural sub-regions are strongly correlated with actual cleaning needs. Preset region identification rules are used to construct structural templates based on global and local boundaries, dividing the global map into easily covered structural sub-regions. Preset decomposition constraints ensure that the division of structural sub-regions meets the requirements of path continuity and coverage efficiency, reducing redundant paths or coverage blind spots caused by blind segmentation. The combined application of feature vertices and structural templates makes map decomposition conform to both physical boundaries and path characteristics, reducing the coverage loss problem caused by rule template mismatch in traditional methods. By minimizing the distance to feature vertices, the compactness of structural sub-region division and path continuity are ensured, reducing the energy consumption of frequent turning or repeated cleaning by the sweeper. Furthermore, the dynamic configuration mechanism of special sub-regions effectively solves the path planning problem of irregular obstacles or complex environments, improving the passability of sweepers in complex scenarios.

[0064] Based on any embodiment of the system in this application, as an optional embodiment, determining the starting point of the structural sub-region based on feature vertices includes the following steps: Based on a preset coverage algorithm, several starting points that satisfy the motion rules corresponding to the coverage algorithm are generated in any structural sub-region. A preset starting point distance constraint is used to determine the starting point of all structural sub-regions based on minimizing the distance between the starting points of adjacent structural sub-regions.

[0065] As can be seen from the above embodiments, a number of candidate starting points satisfying the motion rules within a sub-region are generated through a preset coverage algorithm, ensuring that the cleaning path conforms to the geometric constraints of the algorithm itself. The generation of candidate starting points is directly related to the motion mode of the coverage algorithm, avoiding path breakage or repeated coverage due to improper selection of starting points. Smooth connection of the global path is achieved through preset starting point distance constraints. The selection of starting points simultaneously satisfies the local optimization requirements of the coverage algorithm and the coherence requirements of the global path, forming a multi-objective collaborative optimization mechanism. Through the joint optimization of motion rules and distance constraints, the feasibility of the cleaning path in complex scenarios is ensured.

[0066] Based on any embodiment of the system in this application, as an optional embodiment, a first path for the covered structure sub-region is generated based on the starting point using a covering algorithm, including the following steps: Construct a grid map of all structural sub-regions, determine motion rules according to a preset coverage algorithm, and for each structural sub-region, traverse the grid map from the starting point according to the motion rules to generate a coverage path, wherein the coverage path includes a starting point and an ending point; A predetermined endpoint distance constraint is set between adjacent structural sub-regions. The target point in the coverage path is determined based on the minimum distance between the endpoint of the coverage path and the starting point of the adjacent structural sub-region. The first path is determined based on the starting point, the target point, and the corresponding coverage path.

[0067] As can be seen, through the above embodiments, by constructing a grid map of all structural sub-regions, the continuous space is discretized into computable grid cells, ensuring a high degree of matching between the coverage path and the physical environment, and avoiding coverage blind spots or path conflicts caused by spatial abstraction errors. A preset endpoint distance constraint is implemented, minimizing the physical distance between the endpoint of the coverage path in adjacent structural sub-regions and the starting point of the next region, achieving a smooth transition between regions. The coverage path generation simultaneously satisfies the coverage integrity within sub-regions and the minimum movement distance for inter-regional connections, forming a global-local collaborative optimization mechanism. Through the joint configuration of the starting point, target point, and coverage path, the sweeper's path throughout the cleaning area is ensured to be compact and efficient, avoiding the global path fragmentation problem caused by local optimal selection in traditional methods. This significantly improves the coverage and efficiency of cleaning operations, reduces energy consumption, and enhances the sweeper's path planning adaptability to complex environments.

[0068] Based on any embodiment of the system in this application, as an optional embodiment, determining the nonlinear constraint region of a special sub-region includes the following steps: Point cloud data of a specific sub-region is acquired, the point cloud data is converted to a unified coordinate system, and roadside geometric parameters are extracted in the unified coordinate system to determine the nonlinear constraint region.

[0069] As can be seen, through the above embodiments, the combination of point cloud data and a unified coordinate system significantly improves the geometric description accuracy of nonlinear constraint regions and enhances the modeling accuracy of nonlinear constraint regions in special sub-regions. The extraction and analysis of roadside geometric parameters enable the system to flexibly respond to irregular obstacles or dynamic environmental changes, improving dynamic adaptability. The accurate identification of nonlinear constraint regions effectively improves the working stability of sweepers in complex terrains. The accurate modeling of nonlinear constraint regions significantly reduces the path planning failure rate, improves sweeping efficiency and equipment safety, and is especially suitable for sweeper operations in complex environments. This further improves the automation level and completion quality of sweeping operations, enabling sweepers to adapt to actual application scenarios in complex environments and greatly expanding the application scope.

[0070] Based on any embodiment of the system in this application, as an optional embodiment, the second path of a special sub-region is generated by transforming the nonlinear constraint region based on coordinate system transformation, including the following steps: For the nonlinear constraint region, key points are modeled by transforming the unified coordinate system into the Frenet coordinate system; The number of key points is determined based on the maximum width of the nonlinear constraint region and the width of the sweeper. The key points are then updated through a quadratic coordinate transformation based on the positional order of the key points within the nonlinear constraint region to determine the target key points in a unified coordinate system. A Bézier curve is introduced to smooth the target key points under the nonlinear constraint region to generate a second path for a special sub-region. The second path includes a starting point and a target point. The expression for the key points in Frenet coordinates is: ; ; In the formula, This represents the x-coordinate of the i-th key point. This represents the ordinate of the i-th key point. This represents the coordinates of the left vertex of the nonlinear constraint region. This represents the coordinates of the right vertex of the nonlinear constraint region, where N represents the number of key points. Represents key points in Frenet coordinates. Represents key points in a unified coordinate system; The expression for smoothing the Bézier curve is: ; In the formula, , , , The continuous inflection points represent the control points, and t represents the proportional coefficient.

[0071] As can be seen, through the above embodiments, by transforming the nonlinear constraint region from a unified coordinate system to a Frenet coordinate system, a dynamic parameterized description of complex terrain is achieved. The introduction of the Frenet coordinate system enables key point modeling to be directly related to the curvature characteristics of the path, avoiding the path distortion problem caused by coordinate offset in the traditional Cartesian coordinate system. It can also transform the nonlinear constraint problem under the original unified coordinate system into a linear constraint problem for key point solving. The number of key points is dynamically determined according to the maximum width of the nonlinear constraint region and the width of the sweeper, ensuring that the path coverage matches the physical size of the equipment. The key point positions are updated through a quadratic coordinate transformation to ensure that the key points strictly follow the geometric boundaries of the nonlinear constraint region, avoiding path deviation or conflict. Bézier curves are introduced to the target key points for smoothing, generating a continuous and differentiable second path, significantly reducing the sweeper's sharp turns or frequent steering in complex terrain, allowing the path to flexibly adapt to the shape of the nonlinear constraint region (such as S-shaped bends and U-shaped obstacles), while preserving the connectivity between the starting point and the target point, ensuring the global feasibility of the path. This enables path planning to adapt to nonlinear constraint regions of different shapes, improving the stability and efficiency of the sweeper's operation. Consequently, the sweeper can adapt to real-world application scenarios in complex environments, significantly expanding its application scope.

[0072] Based on any embodiment of the system in this application, as an optional embodiment, the starting point and target point of any path are determined, and the starting point and target point are connected based on an improved path search algorithm to generate a target path, including the following steps: Based on the Hybrid A* algorithm, an evaluation function is set to minimize the cost of connecting the target point to the starting point of adjacent paths. The target path is generated by connecting the starting point and the target point in all first and second paths. The evaluation function of the Hybrid A* algorithm is expressed as follows: ; ; In the formula, This represents the evaluation function of the Hybrid A* algorithm. This represents the actual cumulative cost from the target point to the current starting point n. This represents the heuristic value based on Euclidean distance. This represents the cost of calculating distance using the Reeds-Shepp curve, with max indicating the value of taking... or The maximum value in the heuristic function is used as the cost. The actual cost represents the actual path length from the target point to the current starting point. This indicates the cost of the shift or change. This indicates the cost of changing gears. This indicates the cost of reversing. This indicates the weight of the actual cost. The weights representing the shifts in direction. This indicates the weight of gear shifting. Indicates the weight of reversing. The weight represents the actual cumulative cost of the i-th current starting point n. This represents the actual cumulative cost of the i-th current starting point n.

[0073] As can be seen, through the above embodiments, the improved Hybrid A* algorithm and multi-dimensional cost evaluation function design achieve the coordinated optimization of global optimality and local smoothness of path planning in complex environments, significantly improving the path continuity and movement feasibility of the sweeper when switching between structural sub-regions and special sub-regions. By connecting the starting point and target point of the first path (structural sub-region coverage path) and the second path (special sub-region obstacle avoidance path), a global target path is generated, ensuring seamless transition of the sweeper between regular and complex areas. This significantly reduces ineffective movement and equipment wear in path planning, improves sweeping efficiency and safety in complex environments, and is especially suitable for complex environmental operation scenarios that require frequent switching of movement modes.

[0074] Based on any embodiment of the system in this application, as an optional embodiment, another embodiment of this application provides a sweeper path planning device suitable for complex environments. This sweeper path planning device for complex environments includes a processor, a computer-readable storage medium, a memory, and a network interface connected via a system bus. The computer-readable non-volatile storage medium of this sweeper path planning device for complex environments stores an operating system, a database, and computer-readable instructions. The database may store information sequences. When the computer-readable instructions are executed by the processor, the processor can implement a sweeper path planning method suitable for complex environments.

[0075] The processor of this sweeper path planning device for complex environments provides computational and control capabilities, supporting the operation of the entire device. The memory of this device can store computer-readable instructions, which, when executed by the processor, cause the processor to perform the sweeper path planning method for complex environments as described in this application. The network interface of this device is used for communication with a terminal.

[0076] In this embodiment, the processor executes the specific functions of each module in the system, and the memory stores the program code and various types of data required to execute the aforementioned modules or sub-modules. The network interface is used to enable data transmission between user terminals or servers.

[0077] The non-volatile readable storage medium in this embodiment stores the program code and data required to execute all modules in the sweeper path planning system applicable to complex environments of this application. The server can call the program code and data of the server to execute the functions of all modules.

[0078] This application also provides a non-volatile readable storage medium storing computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to perform the steps of the sweeper path planning method applicable to complex environments according to any embodiment of this application.

[0079] This application also provides a computer program product, including a computer program / instructions that, when executed by one or more processors, implement the steps of the method described in any embodiment of this application.

Claims

1. A path planning method for a sweeper vehicle applicable to complex environments, characterized in that, Includes the following steps: Obtain a global map of the cleaning area, determine feature vertices to decompose the global map, and generate structural sub-regions and special sub-regions; The starting point of the structured sub-region is determined based on the feature vertices, and a first path covering the structured sub-region is generated based on the starting point using a covering algorithm. The nonlinear constraint region of a special sub-region is determined, and the nonlinear constraint region is transformed based on coordinate system transformation to generate a second path for the special sub-region; Determine the starting point and target point of any path, and connect the starting point and target point based on an improved path search algorithm to generate the target path of the cleaning area.

2. The sweeper path planning method applicable to complex environments according to claim 1, characterized in that, Determining feature vertices to decompose the global map and generate structured sub-regions and special sub-regions includes the following steps: Parse the global map to obtain obstacle information and global map boundaries, determine local map boundaries based on obstacle information, and determine feature vertices based on either mandatory points or repeated points on the path in the global map. Preset region identification rules are used to determine the structural template of the decomposed region based on the global map boundary and the local map boundary. The structural template is then used to decompose the global map and generate structural sub-regions. Preset region decomposition constraints are applied to the global map of the structure template decomposition based on minimizing the distance to feature vertices of all decomposed regions. Unstructured sub-regions are configured as special sub-regions based on global map boundaries and path characteristics.

3. The sweeper path planning method applicable to complex environments according to claim 1, characterized in that, Determining the starting point of a structural sub-region based on feature vertices includes the following steps: Based on a preset coverage algorithm, several starting points that satisfy the motion rules corresponding to the coverage algorithm are generated in any structural sub-region. A preset starting point distance constraint is used to determine the starting point of all structural sub-regions based on minimizing the distance between the starting points of adjacent structural sub-regions.

4. The sweeper path planning method applicable to complex environments according to claim 1, characterized in that, The first path of the covered sub-region is generated based on the starting point using a covering algorithm, including the following steps: Construct a grid map of all structural sub-regions, determine motion rules according to a preset coverage algorithm, and for each structural sub-region, traverse the grid map from the starting point according to the motion rules to generate a coverage path, wherein the coverage path includes a starting point and an ending point; A predetermined endpoint distance constraint is set between adjacent structural sub-regions. The target point in the coverage path is determined based on the minimum distance between the endpoint of the coverage path and the starting point of the adjacent structural sub-region. The first path is determined based on the starting point, the target point, and the corresponding coverage path.

5. The sweeper path planning method applicable to complex environments according to claim 1, characterized in that, Determining the nonlinear constraint region of a specific sub-region includes the following steps: Point cloud data of a specific sub-region is acquired, the point cloud data is converted to a unified coordinate system, and roadside geometric parameters are extracted in the unified coordinate system to determine the nonlinear constraint region.

6. The sweeper path planning method applicable to complex environments according to claim 5, characterized in that, Based on coordinate system transformation, a second path is generated for a special sub-region, including the following steps: For the nonlinear constraint region, key points are modeled by transforming the unified coordinate system into the Frenet coordinate system; The number of key points is determined based on the maximum width of the nonlinear constraint region and the width of the sweeper. The key points are then updated through a quadratic coordinate transformation based on the positional order of the key points within the nonlinear constraint region to determine the target key points in a unified coordinate system. A Bézier curve is introduced to smooth the target key points under the nonlinear constraint region to generate a second path for a special sub-region. The second path includes a starting point and a target point. The expression for the key points in Frenet coordinates is: ； ； In the formula, This represents the x-coordinate of the i-th key point. This represents the ordinate of the i-th key point. This represents the coordinates of the left vertex of the nonlinear constraint region. This represents the coordinates of the right vertex of the nonlinear constraint region, where N represents the number of key points. Represents key points in Frenet coordinates. Represents key points in a unified coordinate system; The expression for smoothing the Bézier curve is: ； In the formula, , , , The continuous inflection points represent the control points, and t represents the proportional coefficient.

7. The sweeper path planning method applicable to complex environments according to claim 1, characterized in that, Determine the starting point and target point of any path, and connect the starting point and target point based on an improved path search algorithm to generate the target path, including the following steps: Based on the Hybrid A* algorithm, an evaluation function is set to minimize the cost of connecting the target point to the starting point of adjacent paths. The target path is generated by connecting the starting point and the target point in all first and second paths. The evaluation function of the Hybrid A* algorithm is expressed as follows: ； ； In the formula, This represents the evaluation function of the Hybrid A* algorithm. This represents the actual cumulative cost from the target point to the current starting point n. This represents the heuristic value based on Euclidean distance. This represents the cost of calculating distance using the Reeds-Shepp curve, with max indicating the value of taking... or The maximum value in the heuristic function is used as the cost. The actual cost represents the actual path length from the target point to the current starting point. This indicates the cost of the shift or change. This indicates the cost of changing gears. This indicates the cost of reversing. This indicates the weight of the actual cost. The weights representing the shifts in direction. This indicates the weight of gear shifting. Indicates the weight of reversing. The weight represents the actual cumulative cost of the i-th current starting point n. This represents the actual cumulative cost of the i-th current starting point n.

8. A sweeper path planning system suitable for complex environments, characterized in that, The system is used to execute the sweeper path planning method applicable to complex environments as described in any one of claims 1-7, and the system includes: The global decomposition module is used to obtain a global map of the cleaning area, determine feature vertices to decompose the global map, and generate structural sub-regions and special sub-regions. The first planning module is used to determine the starting point of the structured sub-region based on the feature vertices, and to generate the first path covering the structured sub-region based on the starting point using a covering algorithm. The second planning module is used to determine the nonlinear constraint region of a special sub-region, transform the nonlinear constraint region based on coordinate system transformation, and generate the second path of the special sub-region. The path generation module is used to determine the starting point and target point of any path, and connect the starting point and target point based on an improved path search algorithm to generate the target path of the cleaning area.

9. A sweeper path planning device suitable for complex environments, comprising: At least one processor, and, A memory communicatively connected to the at least one processor; characterized in that the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the sweeper path planning method applicable to complex environments as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores, in the form of computer-readable instructions, a computer program implemented according to any one of claims 1 to 7, which, when invoked by a computer, executes the steps included in the corresponding method.

Images