Layered search three-dimensional airspace path planning method and system based on adaptive resolution grid

Abstract

The invention particularly relates to a hierarchical search three-dimensional airspace path planning method and system based on an adaptive resolution grid. The method comprises the following steps: constructing a corresponding multi-resolution grid model based on a three-dimensional grid map; global path planning is carried out on the coarse resolution grids, coarse paths of all coarse grid nodes are determined, and a coarse path sequence is constructed; performing non-overlapping segmentation on the coarse path sequence; planning local windows matched with the fine-resolution grids, and determining window entrance and exit points corresponding to the local windows; performing obstacle recognition on each local window based on the fine resolution grid, calling a corresponding path generation strategy according to an obstacle recognition result of the local window, and determining a local refined sub-path corresponding to each local window; and splicing the path information of each local window in sequence to obtain a path planning result. According to the method, efficient and reliable path planning of a large-scale and complex three-dimensional airspace environment can be realized.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) path planning technology, specifically to a hierarchical search method and system for three-dimensional airspace path planning based on adaptive resolution grids. Background Technology

[0002] In related technologies, unmanned aerial vehicles (UAVs) are required to perform autonomous path planning tasks in complex three-dimensional environments for applications such as low-altitude logistics, urban inspection, emergency rescue, and military reconnaissance. These environments typically have the following characteristics: a large airspace and a vast number of three-dimensional spatial states; complex obstacle types, including terrain, buildings, no-fly zones, and temporary restricted areas; highly irregular obstacle distribution; and simultaneous requirements for path safety, smoothness, and real-time performance.

[0003] Traditional A Algorithms typically perform path search based on a uniform resolution grid. When the grid resolution is high, the number of search nodes increases dramatically, leading to excessive computational complexity. When the grid resolution is low, path accuracy is insufficient, and safety hazards such as path crossing obstacles or getting too close to obstacles are easily encountered. Furthermore, performing a uniform grid search directly in a 3D environment presents the following problems: uniform grid discretization leads to an exponential increase in the state space, causing the "curse of dimensionality," and a sharp rise in computational complexity, making it difficult to meet real-time planning requirements; in complex and constrained environments, high-density obstacles and irregular "dead zones" can easily cause the algorithm to get stuck in local optima or path deadlock, and the generated path contains many polyline inflection points, resulting in poor flight adaptability; fixed-resolution grid modeling cannot balance planning accuracy and computational efficiency. While coarse-resolution grids can speed up the search, they cannot guarantee path safety, while fine-resolution grids can improve planning accuracy but significantly increase computational overhead.

[0004] While existing multi-resolution path planning methods alleviate the computational burden of uniform resolution grids in large-scale spaces to some extent, they still suffer from several key shortcomings in engineering applications. First, the path connection mechanism between different resolution levels is imperfect, and the multi-resolution switching process lacks a stable and unified transition strategy, easily leading to discontinuities or offsets between coarse and fine paths, thus affecting the feasibility and consistency of the overall path. Second, the local refinement process often occurs independently of the global path structure, lacking effective guidance from the global skeleton path, making local searches prone to deviating from the global optimum, or even getting stuck in local optima or failing. Third, in complex, obstacle-dense, or spatially constrained areas, existing methods are prone to path breaks, local unsolvable problems, or frequent backtracking, resulting in insufficient stability and robustness of the overall planning. Furthermore, existing multi-resolution methods lack unified and standardized rules for constructing and selecting local planning windows; window size and position often rely on empirical settings, making it difficult to balance search efficiency and path accuracy, affecting the algorithm's adaptability and consistency in different scenarios.

[0005] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of the present invention, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] This invention provides a hierarchical search method for three-dimensional spatial path planning based on an adaptive resolution grid, and a hierarchical search system for three-dimensional spatial path planning based on an adaptive resolution grid. This method can achieve efficient and reliable path planning for large-scale and complex three-dimensional spatial environments, thereby overcoming the shortcomings of existing technologies to a certain extent.

[0007] Other features and advantages of the invention will become apparent from the following detailed description, or may be learned in part by practice of the invention.

[0008] According to a first aspect of the present invention, a hierarchical search method for three-dimensional spatial path planning based on an adaptive resolution grid is provided, the method comprising: A multi-resolution mesh model is constructed based on a 3D raster map; the multi-resolution mesh model includes: coarse-resolution mesh and fine-resolution mesh; Based on the start and end point information corresponding to the current path planning task, global path planning is performed on the coarse resolution grid to determine the coarse path of each coarse grid node and construct the coarse path sequence. The coarse path sequence is segmented into non-overlapping segments; local windows adapted to fine-resolution grids are planned, and the entry and exit points of each local window are determined. Obstacles are identified in each local window based on a fine-resolution mesh, and the corresponding path generation strategy is called based on the obstacle identification results of the local window to determine the local refined sub-path corresponding to each local window. The path information of each local window is sequentially spliced ​​together to obtain the path planning result; the path information of the local window includes local fine-grained sub-paths or coarse-resolution paths.

[0009] In some exemplary embodiments, constructing a corresponding multi-resolution mesh model based on a 3D raster map includes: The 3D raster map is downsampled in the horizontal direction based on a preset downsampling factor, and a coarse-resolution grid is constructed based on the sampling results. The coarse-resolution grid is downsampled horizontally based on a preset downsampling factor, and a fine-resolution grid is constructed based on the sampling results; and The projection of the obstacle onto the horizontal plane is expanded based on a preset size to obtain the obstacle mask under coarse resolution and fine resolution grids; The downsampling factor is configured based on the current terrain features corresponding to the three-dimensional raster map.

[0010] In some exemplary embodiments, global path planning is performed on the coarse-resolution grid based on the start and end point information corresponding to the current path planning task to determine the coarse path sequence, including: Based on the coarse-resolution grid occupancy mask, the improved A is called. The method solves the globally connected path between the start and end points and outputs a coarse path sequence; among them, the improved A The methods include: f(g m )=g (g m )+α・h (g m ) Where g(gm) is the cumulative Euclidean distance cost from the starting point to the current coarse grid gm, h(gm) is the Euclidean distance heuristic cost from the current coarse grid gm to the endpoint, and α is the heuristic weight.

[0011] In some exemplary embodiments, the coarse path sequence is segmented into non-overlapping segments; and local windows adapted to the fine-resolution mesh are planned, and the entry and exit points of each local window are determined, including: Configure the size of the local window based on the coarse grid node data in the coarse path sequence; Based on the size of the local window, the coarse path sequence is segmented into non-overlapping segments to obtain multiple groups of non-overlapping coarse path nodes. For each coarse path node, configure its first minimum bounding rectangle on the horizontal plane in the coarse resolution grid; and map the first minimum bounding rectangle to the fine resolution grid to determine the second minimum bounding rectangle on the fine resolution grid, so as to obtain the pixel index range of the local window based on the fine resolution grid; Based on the pixel index range of the local window using a fine-resolution grid, the preset priority rules are invoked to configure the entry and exit points of each local window.

[0012] In some exemplary embodiments, Configure the entry and exit points of each local window by invoking preset priority rules, including: The priority rules for selecting entry points include the following from highest to lowest priority: the starting point corresponding to the current path planning task, the exit point of the previous local window, the intersection of the skeleton path edge and the local window boundary, and the coarse resolution grid node. The priority rules for selecting exit points include the following from highest to lowest priority: the endpoint corresponding to the current path planning task, the entry point of the next local window, the intersection of the skeleton path edge and the local window boundary, and the coarse-resolution grid node.

[0013] In some exemplary embodiments, obstacle recognition is performed on each local window based on a fine-resolution mesh, and the corresponding path generation strategy is invoked according to the obstacle recognition results of the local windows to determine the local refined sub-path corresponding to each local window, including: Identify whether there are any occupied grid cells within the current local window; and if it is determined that there are occupied grid cells, configure the current local window as a window containing obstacles; For windows containing obstacles, a fine-mesh search domain is configured with the entry and exit points as the start and end points; and an improved A function is invoked for the fine-mesh search domain. The method performs a global connected path search on the start and end points, generates continuous path segments based on a fine-resolution grid, and configures them as locally refined sub-paths. Alternatively, for an accessible window, configure the coarse path node corresponding to the local window as a continuous path segment corresponding to the accessible window, and configure it as a local refined sub-path.

[0014] In some exemplary embodiments, for windows containing obstacles, a fine-mesh search domain is configured with the entry and exit points as the start and end points, including: Based on the coordinate information of the entry and exit points under fine-resolution mesh, the minimum bounding rectangle of the horizontal plane is generated; wherein, the minimum bounding rectangle is determined based on the extreme values ​​of the coordinates of the entry and exit points; The minimum bounding rectangle of the horizontal plane is superimposed with a height range to form an axis-aligned bounding box, which is then configured as the base search region based on a fine-resolution grid.

[0015] In some exemplary embodiments, the method further includes: Based on the expansion parameter S, the basic search region is uniformly expanded outward in the horizontal direction, and the expanded region is configured as the fine mesh search domain.

[0016] In some exemplary embodiments, the method further includes: When the aircraft moves based on the path planning results, identify the diagonal grid of the current grid or the adjacent grids of each axis of the spatial diagonal; If any adjacent grid along an axis contains an obstacle, movement to that diagonal grid or spatial diagonal is prohibited.

[0017] According to a second aspect of the present invention, a hierarchical search three-dimensional spatial path planning system based on an adaptive resolution grid is provided, the system comprising: The multi-resolution mesh model creation module is used to construct corresponding multi-resolution mesh models based on 3D raster maps; the multi-resolution mesh models include: coarse-resolution meshes and fine-resolution meshes; The coarse path sequence calculation module is used to perform global path planning on the coarse resolution grid based on the start and end point information corresponding to the current path planning task, determine the coarse path of each coarse grid node, and construct the coarse path sequence. The local window calculation module is used to perform non-overlapping segmentation of the coarse path sequence; and to plan local windows that adapt to the fine resolution grid, and determine the window entry and exit points corresponding to each local window; The local fine-grained sub-path calculation module is used to identify obstacles in each local window based on a fine-resolution grid, and to call the corresponding path generation strategy based on the obstacle identification results of the local window to determine the local fine-grained sub-path corresponding to each local window. The path planning generation module is used to sequentially splice the path information of each local window to obtain the path planning result; the path information of the local window includes local fine-grained sub-paths or coarse-resolution paths.

[0018] According to a third aspect of the present invention, a computer program product is provided, on which a computer program is stored, wherein when the computer program is executed by a processor, the above-described hierarchical search three-dimensional spatial path planning method based on adaptive resolution grid is implemented.

[0019] According to a fourth aspect of the present invention, a storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the above-described hierarchical search three-dimensional spatial path planning method based on an adaptive resolution grid.

[0020] According to a fifth aspect of the present invention, an unmanned aerial vehicle (UAV) is provided, comprising: The main body of the drone; and the processor and memory installed on the main body of the drone; The memory is used to store the executable instructions of the processor; the processor is configured to implement the above-described hierarchical search three-dimensional spatial path planning method based on adaptive resolution grid when executing instructions via the executable instructions.

[0021] The hierarchical search 3D airspace path planning method based on adaptive resolution grids provided in this invention first obtains a coarse path sequence through global path planning on a coarse resolution grid; then, it refines sub-paths locally on a fine resolution grid; and finally, it stitches the path information to obtain the path planning result, thus realizing a two-level hierarchical path planning strategy. Simultaneously, it effectively reduces the number of inflection points on the path polygons, significantly lowering the average turning angle compared to methods using only fine resolution grids, reducing the cost of UAV attitude adjustment, and improving flight adaptability. By refining the local obstacle-containing areas based on a fine resolution grid, the high computational complexity of planning with a single fine resolution grid is avoided, and the planning time is effectively reduced. This method effectively achieves multi-objective optimization of computational efficiency, path smoothness, safety gap, and airspace utilization, improving the path planning performance of UAVs in complex low-altitude environments.

[0022] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the invention. Attached Figure Description

[0023] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention. It is obvious that the drawings described below are merely some embodiments of the invention, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0024] Figure 1 The illustration shows a schematic diagram of a hierarchical search three-dimensional spatial path planning method based on an adaptive resolution grid, as an exemplary embodiment of the present invention. Figure 2 This schematic diagram illustrates an XY-plane projection dilatation according to an exemplary embodiment of the present invention. Figure 3 The diagram illustrates the principle of a corner constraint mechanism according to an exemplary embodiment of the present invention. Figure 4 This diagram schematically illustrates the effect of adaptive resolution grid route planning according to an exemplary embodiment of the present invention. Figures 5a-5d This diagram illustrates a route planning result with different origin and destination points, as per an exemplary embodiment of the present invention. Figure 6 This schematic diagram illustrates a three-dimensional effect of route planning according to an exemplary embodiment of the present invention. Figure 7 The diagram illustrates an exemplary embodiment of the present invention: a hierarchical search three-dimensional spatial path planning system based on an adaptive resolution grid. Detailed Implementation

[0025] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the invention will be more comprehensive and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0026] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0027] In related technologies, existing multi-resolution path planning methods still generally suffer from several key shortcomings in engineering applications. First, the path connection mechanism between different resolution levels is imperfect, and the multi-resolution switching process lacks a stable and unified transition strategy, easily leading to discontinuities or deviations between coarse and fine paths, thus affecting the feasibility and coherence of the overall path. Second, the local refinement process often occurs independently of the global path structure, lacking effective guidance from the global skeleton path, making local searches prone to deviating from the global optimum, or even getting stuck in local optima or failing. Third, in complex, obstacle-dense, or spatially constrained areas, existing methods are prone to path breakage, local unsolvable problems, or frequent backtracking, resulting in insufficient stability and robustness of the overall planning. Furthermore, existing multi-resolution methods lack unified and standardized rules for constructing and selecting local planning windows; window size and position often rely on empirical settings, making it difficult to balance search efficiency and path accuracy, affecting the algorithm's adaptability and consistency in different scenarios.

[0028] To address the shortcomings and deficiencies of existing technologies, this example implementation provides a hierarchical search method for three-dimensional spatial path planning based on adaptive resolution grids. (Reference) Figure 1 As shown, the method may include the following steps: Step S11: Construct a corresponding multi-resolution grid model based on the 3D raster map; wherein, the multi-resolution grid model includes: coarse resolution grid and fine resolution grid; Step S12: Based on the start and end point information corresponding to the current path planning task, perform global path planning on the coarse resolution grid, determine the coarse path of each coarse grid node, and construct the coarse path sequence. Step S13: Perform non-overlapping segmentation on the coarse path sequence; plan local windows to adapt to the fine resolution grid, and determine the window entry and exit points corresponding to each local window; Step S14: Based on the fine resolution grid, perform obstacle recognition on each local window, and call the corresponding path generation strategy according to the obstacle recognition results of the local window to determine the local refined sub-path corresponding to each local window. Step S15: The path information of each local window is sequentially spliced ​​to obtain the path planning result; wherein, the path information of the local window includes local refined sub-paths or coarse resolution paths.

[0029] The following will describe in more detail each step of the hierarchical search three-dimensional spatial path planning method based on adaptive resolution grid in this exemplary embodiment, with reference to the accompanying drawings and embodiments.

[0030] In step S11, a corresponding multi-resolution grid model is constructed based on the three-dimensional raster map; wherein, the multi-resolution grid model includes: coarse resolution grid and fine resolution grid.

[0031] For example, in step S11, constructing a corresponding multi-resolution grid model based on the 3D raster map includes: Step S21: The three-dimensional raster map is downsampled in the horizontal direction based on a preset downsampling factor, and a coarse resolution grid is constructed based on the sampling results. Step S22: Downsample the coarse-resolution grid in the horizontal direction based on a preset downsampling factor, and construct a fine-resolution grid based on the sampling results; and Step S23: Perform dilation processing on the projection of the obstacle on the horizontal plane based on a preset size to obtain the obstacle mask under coarse resolution grid and fine resolution grid.

[0032] The downsampling factor is configured based on the current terrain features corresponding to the three-dimensional raster map.

[0033] Specifically, for the task of planning the path of a drone, a multi-resolution grid model can be pre-constructed based on a 3D raster map. Specifically, the 3D raster map is denoted as ℳ∈{0,1}^(H×W×Z); where ℳ(i,j,k)=1, indicating that cells (i,j,k) are occupied by obstacles; H and W represent the length and width of the XY plane (horizontal direction), respectively; and Z represents the height or number of layers.

[0034] For 3D raster maps, downsampling can be performed only in the XY plane by a preset downsampling factor f to construct a coarse-resolution grid. This coarse-resolution grid is used for fast global search. Based on the coarse-resolution grid, further downsampling is performed to generate a fine-resolution grid. This fine-resolution grid is only executed in local subdomains to complete feasibility verification and high-quality trajectory generation. The size of the fine-resolution grid can be configured as (H / f) × (W / f) × Z. The fine-resolution grid only downsamples in the XY directions; any cell occupying the Z direction is considered part of the coarse grid.

[0035] For example, the downsampling factor f can be configured based on the requirements of global search efficiency and global path guidance accuracy. Specifically, the downsampling factor f needs to balance global search efficiency (the larger f is, the fewer search nodes are needed, resulting in higher efficiency) and global path guidance accuracy (the larger f is, the lower the coarse grid resolution, resulting in lower guidance accuracy). Specifically, the value of the downsampling factor f can be configured based on the planning area scale, global obstacle density, and onboard computing power.

[0036] Additionally, refer to Figure 2 As shown, a spherical dilation operation with a radius of r (in grid units) can also be applied to the obstacle projection result in the XY plane to obtain a two-dimensional dilated projection D_xy=dilate (M_xy;r). D_xy is then broadcast to all height layers Z in the three-dimensional space to construct a conservative obstacle mask M'. This conservative mask is used for downsampling processing and coarse grid path planning. By expanding the effective range of the obstacle, the collision detection error caused by the quantization and discretization in the three-dimensional space is eliminated.

[0037] In step S12, based on the start and end point information corresponding to the current path planning task, global path planning is performed on the coarse resolution grid to determine the coarse path of each coarse grid node and construct a coarse path sequence.

[0038] For example, in step S12 above, global path planning is performed on the coarse-resolution grid based on the start and end point information corresponding to the current path planning task to determine the coarse path sequence, including: Based on the coarse-resolution grid occupancy mask, the improved A is called. The method solves the global connected path between the start and end points and outputs a coarse path sequence.

[0039] Specifically, for the current path planning task of the UAV, after determining the start and end points, a first-level global path planning process can be performed based on a coarse-resolution grid. Specifically, this can be based on a coarse-resolution grid occupancy mask, using an improved A... The algorithm quickly solves for globally connected paths from the starting point to the ending point. The evaluation function of the improved Algorithm A is defined as: f(gm )=g (g m )+α・h (g m ) Where, g (g m () is the starting point to the current coarse grid g m The cumulative Euclidean distance actual cost, h(g) m ) represents the current coarse raster g m The Euclidean distance to the destination is the heuristic cost; α is the heuristic weight. By balancing the actual cost and the heuristic cost, the global optimality of the path is guaranteed while reducing the size of the search space.

[0040] The coarse path sequence obtained by solving can be represented as Pc={g1,g2,…,g m}, where g m Represented as the m-th coarse grid node, this coarse path sequence is used to quickly determine the connectivity between the starting point and the ending point, as well as the approximate path corridor, and can be used to provide a global guiding framework for subsequent local fine grid refinement.

[0041] In step S13, the coarse path sequence is segmented into non-overlapping segments; local windows adapted to fine resolution grids are planned, and the window entry and exit points corresponding to each local window are determined.

[0042] For example, in step S13 above, the coarse path sequence is segmented into non-overlapping segments; and local windows adapted to the fine-resolution mesh are planned, and the entry and exit points of each local window are determined, including: Step S31: Configure the size of the local window based on the coarse grid node data in the coarse path sequence; Step S32: Perform non-overlapping segmentation on the coarse path sequence based on the size of the local window to obtain multiple non-overlapping coarse path node groups; Step S33: Configure the first minimum bounding rectangle of each coarse path node on the horizontal plane in the coarse resolution grid; and map the first minimum bounding rectangle to the fine resolution grid to determine the second minimum bounding rectangle on the fine resolution grid, so as to obtain the pixel index range of the local window based on the fine resolution grid. Step S34: Based on the pixel index range of the local window based on the fine resolution grid, call the preset priority rules to configure the entry point and exit point of each local window.

[0043] For example, the entry and exit points of each local window are configured by calling preset priority rules, including: The priority rules for selecting entry points include the following from highest to lowest priority: the starting point corresponding to the current path planning task, the exit point of the previous local window, the intersection of the skeleton path edge and the local window boundary, and the coarse resolution grid node. The priority rules for selecting exit points include the following from highest to lowest priority: the endpoint corresponding to the current path planning task, the entry point of the next local window, the intersection of the skeleton path edge and the local window boundary, and the coarse-resolution grid node.

[0044] Specifically, after obtaining the coarse path sequence, a path index grouping strategy can be used to perform non-overlapping segmentation of the coarse path, generate a local planning window adapted to the fine resolution grid, and adaptively determine the window entry and exit points.

[0045] Specifically, the local window size parameter W can be configured first. The size of the local window can be proportionally configured based on the number of coarse grid nodes in the coarse path sequence. The index interval of the i-th group of coarse path nodes is L=[i×W,min ((i+1)×W-1,M)], i=0,1,…,[M / W]. This interval covers at most W adjacent coarse path nodes. The last group is truncated according to the actual coarse path length, ensuring that the groups do not overlap and completely cover the entire coarse path.

[0046] For the i-th coarse path node, calculate its minimum bounding rectangle on the XY plane, map the bounding rectangle of the coarse resolution grid to the fine resolution grid, and obtain the pixel index range that fits the local fine resolution grid window.

[0047] After completing the local window planning, the entry and exit points for each local window can be solved. Among them, the entry point e is used for... i The priority rules for solving the problem include the following from highest to lowest priority: the starting point of the current path planning task, the exit point of the previous local window, the intersection of the skeleton path edge and the local window boundary, and the coarse-resolution mesh node. Specific logic may include: 1) If the i-th group is the 0th group (i=0, i.e., containing the starting point corresponding to the current path planning task), then directly set the actual starting point (the corresponding fine-resolution grid coordinates are S). f (e0) serves as the entry point; 2) If the i-th group is an intermediate group (i≠0, and i≠[M / W]), and the exit point x{i-1} of the previous group window (the (i-1)-th group) has been determined, then x{i-1} is taken as the entry point e of the current local window. i ; 3) If neither of the above two conditions is met, take the line segment p0p1 on the coarse path skeleton that is adjacent to the front end of the current window, and calculate the intersection point p of this line segment and the rectangular boundary of the current window. int If line segment p0p1 intersects the boundary of the window rectangle at multiple points (in special cases, the line segment crosses the boundary twice), then the distance between each intersection point and p0 is calculated using the Euclidean distance formula, and the intersection point with the smallest distance is selected as the candidate entry point; where the Euclidean distance formula is expressed as: 4) If line segments p0 and p1 do not intersect with the boundary of the window rectangle (e.g., both p0 and p1 are within the window), then p1 (the first node of the current group of coarse path nodes) will be used as the entry point e. i .

[0048] Among them, the exit point x i The priority rules for solving the problem include the following from highest to lowest priority: the endpoint of the current path planning task, the entry point of the next local window, the intersection of the skeleton path edge and the local window boundary, and coarse-resolution mesh nodes. Specifically, this may include: 1) If the i-th group is the last group (i=[M / W], i.e., it contains the endpoint corresponding to the current path planning task), then directly set the global endpoint (corresponding to the fine-resolution grid coordinates G). f x as the exit point i ; 2) If the i-th group is an intermediate group (i≠0 and i≠[M / W]), first pre-calculate the entry point e_{i+1} of the next adjacent group window (the (i+1)-th group), and use e_{i+1} as the exit point x of the current group. i ; 3) If neither of the above two conditions is met, take the line segment p on the coarse path skeleton that is adjacent to the back end of the current window. k p k+1 Calculate the intersection point p_out of the line segment and the current window rectangle boundary; if the line segment p k p k+1 There are multiple intersection points with the window rectangle boundary. The distance between each intersection point and pk+1 is calculated using the Euclidean distance formula, and the intersection point with the smallest distance is selected as the candidate exit point. The Euclidean distance formula is expressed as: 4) If line segment p k p k+1 It has no intersection with the boundary of the window rectangle (e.g., p). k p k+1 If all nodes are within the window, then p_k (the last node of the current group's coarse path nodes) will be used as the exit point.

[0049] In step S14, obstacles are identified for each local window based on the fine-resolution grid, and the corresponding path generation strategy is called according to the obstacle identification results of the local window to determine the local refined sub-path corresponding to each local window.

[0050] For example, step S14 described above may specifically include: Step S41: Identify whether there are any occupied grid cells within the current local window; and if it is determined that there are occupied grid cells, configure the current local window as a window with obstacles; Step S42: For windows containing obstacles, configure a fine-mesh search domain with the entry and exit points as the start and end points; and call the improved A function on the fine-mesh search domain. The method performs a global connected path search on the start and end points, generates continuous path segments based on a fine-resolution grid, and configures them as locally refined sub-paths. Step S43, or, for the accessibility window, configure the coarse path node corresponding to the local window as a continuous path segment corresponding to the accessibility window, and configure it as a local refined sub-path.

[0051] For example, for a window containing obstacles, a fine-grid search domain is configured with the entry and exit points as the starting and ending points, including: Step S51: Generate the minimum bounding rectangle of the horizontal plane based on the coordinate information of the entry point and the exit point under the fine resolution grid; wherein, the minimum bounding rectangle is determined based on the extreme values ​​of the coordinates of the entry point and the exit point. Step S52: Overlay the height range of the minimum outer rectangle of the horizontal plane to form an axis-aligned bounding box, and configure it as the basic search region based on a fine-resolution grid.

[0052] For example, the above method may further include: Step S53: Based on the expansion parameter S, the basic search region is uniformly expanded outward in the horizontal direction, and the expanded region is configured as the fine mesh search domain.

[0053] Specifically, for each local window, it is possible to determine whether there are obstacles within the window, and then plan a local refined sub-path for the window based on the distribution of obstacles within the window.

[0054] Specifically, when determining whether a local window is an obstruction window, this can be done by inspecting window W. f (i) Determine whether there is an occupied grid (ℳ(i,j,k)=1). If there is an occupied grid, it is determined to be a window with obstacles; otherwise, it is determined to be a window without obstacles.

[0055] Specifically, for windows containing obstacles, when planning the corresponding locally refined sub-paths, a search domain based on a fine-resolution grid can be determined, with the entry point e_i and the exit point x_i as the starting and ending points. Restricted A* operations can then be performed within this search domain. The algorithm generates continuous path segments with fine mesh resolution and configures these continuous path segments as locally refined sub-paths within a local window. Correspondingly, constraints for path planning within the window can also be configured, including: the path must be within the window and maintain a minimum safe distance from obstacles. The planned continuous path segment contains multiple continuous fine mesh nodes, denoted as P_i={p_i1, p_i2,..., p_in_i}, where p_i1=e_i and p_in_i=x_i.

[0056] Specifically, when configuring the search domain, the local refinement boundary can be controlled via the parameter 's'. For example, if it is necessary to connect two representative points in a fine-resolution mesh, the search domain can be extended by several cells on each side of the axis-aligned bounding rectangle of these two points. This can include the following rules: Suppose we need to connect two representative points (A and B) in a fine-resolution mesh, namely the exit point and the entry point. Based on the coordinates of these two points in the fine-resolution mesh, we generate a minimum bounding rectangle in the top-view plane. In the top-view two-dimensional plane (XY direction, corresponding to the UAV's horizontal flight plane), this rectangle is determined by the extreme values ​​of the coordinates of the two representative points. The formula can be expressed as: x_min = min(x-coordinate of A, x-coordinate of B) x_max = max(x-coordinate of A, x-coordinate of B) y_min = min(y-coordinate of A, y-coordinate of B) y_max = max(y coordinate of A, y coordinate of B) In three-dimensional space, a height range (Z direction) is superimposed to form an axis-aligned bounding box, corresponding to the basic search region of the fine mesh. In some embodiments, this basic search region can be directly configured as the fine mesh search domain for global connected path search.

[0057] Further optimization can utilize the parameter 's' to uniformly expand the four sides (left, right, top, and bottom, corresponding to the XY plane) of the rectangle aligned to the base axis, with each side expanding outwards by 's' finer mesh cells. The coordinate range of the expanded fine-resolution mesh search area is then defined. The corresponding formula can be expressed as: After expansion, x_min' = x_min - s (expanding by s cells on the left). After expansion, x_max' = x_max + s (the right side expands by s cells). After expansion, y_min' = y_min - s (s cells are expanded on the lower side). After expansion, y_max' = y_max + s (expanding the upper side by s cells) For example, the specific value of the extended parameter s can be configured using a tiered method, such as setting it according to the obstacle density ρ_obst within the local window. For instance, when ρ_obst < 0.2, s = 1~2; when 0.2 ≤ ρ_obst ≤ 0.5, s = 3~4; and when ρ_obst > 0.5, s = 5~6. In three-dimensional space, to reduce the search burden, the algorithm of this invention mainly performs layered search in the top view plane, so it does not expand in the height direction, but still superimposes the original height range to form a new axis-aligned bounding box.

[0058] Specifically, for an accessible window, since the window does not contain any obstacles, there is no need to generate a fine-resolution path. Instead, the coarse path nodes of the current group are directly extracted and used as the path segment P_i corresponding to the window.

[0059] In step S15, the path information of each local window is spliced ​​sequentially to obtain the path planning result; wherein, the path information of the local window includes local refined sub-paths or coarse resolution paths.

[0060] For example, refer to Figure 4 , Figure 6 As shown, after obtaining the coarse path sequence and the local refined sub-paths corresponding to each local window, all the local refined sub-paths and the coarse resolution path segments that do not need refinement can be concatenated in sequence. All path segments are arranged in ascending order of group index i, namely P0 (the 0th group of path segments), P1 (the 1st group of path segments), ..., P_n-1 (the last group of path segments), to obtain the initial concatenated path P_init=P1∪P2∪...∪P_i∪...∪P_n.

[0061] For example, the method further includes: when the aircraft is moving based on the path planning result, identifying each axial adjacent cell of the current grid's diagonal grid or spatial diagonal; if any axial adjacent cell contains an obstacle, then moving to that diagonal grid or spatial diagonal is prohibited.

[0062] Specifically, refer to Figure 3As shown, a neighborhood model and corner constraints can be pre-configured. Specifically, these constraints can include: based on a highly connected neighborhood model, when an aircraft intends to move from the current grid (x, y, z) to a diagonal grid (planar diagonal (x+1, y+1, z) or spatial diagonal (x+1, y+1, z+1)), it first checks whether all axial neighboring grids corresponding to the diagonal movement (two adjacent axial grids forming the diagonal path) are in an obstacle state. If any axial neighboring grid is occupied by an obstacle, the diagonal movement is prohibited to avoid the collision risk caused by diagonal crossing. Specifically, the corner constraint can be pre-configured as a coarse path planning constraint when constructing a coarse path sequence based on a coarse-resolution grid. It can also be pre-configured as a fine-grained sub-path planning constraint when planning a local fine-grained sub-path for a local window based on a fine-resolution grid.

[0063] In some exemplary embodiments, this constraint can also be configured as a constraint on the actual path planning result executed by the aircraft. For example, when the UAV executes the path planning result, obstacles can be judged in advance based on this constraint on the planned trajectory to be executed, thereby further realizing effective obstacle avoidance.

[0064] In one exemplary embodiment, the following environment can be configured to verify the method of the present invention.

[0065] The configuration of geographic environment data includes: using real terrain data of a certain area, obtaining terrain data through a digital elevation model (DEM), and using an interpolation algorithm to generate a ground elevation field, representing the ground elevation at any horizontal coordinate (x,y).

[0066] The configuration of obstacles includes: arbitrarily defining circular and polygonal restricted airspace (allowing overlap). Obstacle areas such as buildings and restricted areas are divided into polygonal obstacles (defined by the set of vertices to define the horizontal projection boundary) and circular obstacles (defined by the center coordinates and radius to define the horizontal range). Both types of obstacles specify a height range.

[0067] The grid configuration parameters include: a downsampling factor of f=4 for the coarse grid, meaning each coarse grid cell equals 4×4=16 fine grid cells; an obstacle projection expansion radius r=3 (in grid units); a local window size parameter W=3; a local refinement parameter s=6, meaning the maximum and minimum values ​​of the coarse resolution grid points in this window are used as the local region search boundary; and a heuristic weight α=1, using equivalent weights.

[0068] The configuration comparison algorithms include: selecting the coarse-only A algorithm and the fine-only A algorithm as comparison objects. The coarse-only A algorithm has the same coarse-resolution mesh parameters as the present invention, and the fine-only A algorithm has the same fine-resolution mesh parameters as the present invention.

[0069] Specifically, the process includes the following steps: 1) Flight environment modeling: The occupied voxel method is used to mark the terrain and obstacles in the 3D grid map. The 3D environment is discretized into uniform fine resolution grids. For each grid, it is determined whether its (x,y) coordinates fall within the horizontal projection of the obstacle and whether its height falls within the height range of the obstacle or is lower than the terrain height. If so, it is marked as occupied (ℳ(i,j,k)=1). Obstacle projection dilation processing: Extract the projection M_xy of the fine-resolution mesh obstacle mask onto the XY plane; A morphological dilation operation is performed on M_xy using a circular structuring element (radius r of fine-resolution mesh elements) to obtain a two-dimensional dilated projection D_xy = dilate(M_xy;r); Broadcast D_xy to all height layers Z in 3D space to construct a conservative barrier mask M', where M'(i,j,k)=D_xy(i,j), which is used for subsequent coarse grid downsampling; The conservative barrier mask M' is downsampled. If the downsampling factor f=4, that is, every 4×4 fine grids in the XY direction are merged into 1 coarse grid, while the Z direction remains unchanged. Coarse grid occupancy judgment rule: If at least one fine grid in the 4×4×1 fine grids is occupied (M'(i,j,k)=1), then the corresponding coarse grid is marked as occupied (ℳ_c (i_c,j_c,k)=1), where i_c=[i / 4], j_c=[j / 4]; Finally, a coarse-grid obstacle mask ℳ_c is generated, completing the construction of the multi-precision mesh model.

[0070] 2) Global coarse-grid programming: Taking the start and end points as input, and running an improved A algorithm based on the conservative obstacle mask ℳ_c. The algorithm generates a global coarse path sequence P_c; it should be noted that in A The algorithm uses corner constraint checks during the search process: when a neighboring node m is the diagonal grid (planar diagonal or spatial diagonal) of the current node n, it checks whether the two axial neighboring cells that form the diagonal path are occupied. If either axial neighboring cell is occupied, the expansion of the diagonal node is prohibited.

[0071] 3) Local window partitioning and entry / exit point determination: Local windows are partitioned according to a path index grouping strategy. For example, if the window size parameter W=3 and P_c contains 15 path points, then the grouping index i=0,1,…,4, and the node range of each group is as follows: Group 0 (i=0): Index interval L0=[0,2], containing node P_c (0)={g1,g2,g3} (index 0 corresponds to g1, index 2 corresponds to g3); Group 1 (i=1): Index interval L1=[3,5], containing node P_c (1)={g4,g5,g6}; Group 2 (i=2): Index interval L2=[6,8], containing node P_c (2)={g7,g8,g9}; And so on, Group 5 (i=5): Index interval L5=[12,14], containing node P_c (5)={g13,g14,g15} (end point group); Calculate the minimum bounding rectangle of each index group window and map it to a fine mesh. Adaptively calculate and determine the entry point e_i and exit point x_i of each window according to the priority rule.

[0072] 4) Local fine-mesh path refinement: Within a local window containing obstacles, using e_i and x_i as start and end points, expand by s meshes to define the search domain, running under restricted A... The algorithm generates fine-grained sub-paths, and the search also follows the corner constraint mechanism. Window paths without obstacles continue to retain the original set of coarse-path nodes.

[0073] 5) Path concatenation and post-processing: Arrange the path segments according to the group index order, in the order of P0→P1→P2→P3→P4→P5→P6, concatenate the coarse path segments and fine sub-paths, and output the final path.

[0074] Specifically, refer to Figure 4 The image shows the path planning results obtained using the algorithm of this invention. It can be seen that compared with the single-resolution grid planning method, the adaptive resolution hierarchical processing strategy of this invention exhibits better adaptability in complex obstacle environments: when the path needs to traverse densely populated areas of surrounding obstacles or cope with path bends caused by irregular obstacle distribution, the planned path can flexibly conform to the obstacle distribution contour through the connectivity guidance provided by the global coarse grid and the fine obstacle avoidance adjustment by the local fine grid. Under the premise of avoiding collision risks, it effectively reduces unnecessary path bends, ensuring the safety and feasibility of the path, and improving the smoothness and environmental adaptability of the path. This fully demonstrates the advantages of the hierarchical strategy in balancing path flexibility and obstacle avoidance accuracy.

[0075] Subsequently, several sets of starting and ending points were randomly generated, and three algorithms were used to calculate the flight path. Six core indicators were recorded for each algorithm: average time consumption, number of extended nodes, path length, turning angle, and airspace occupancy, for performance evaluation. Figure 5 shows the path planning results of four sets obtained by the three algorithms. It can be seen that the experimental results show that the algorithm proposed in this invention significantly outperforms the comparative algorithms in terms of overall performance, as shown in Table 1.

[0076] Table 1 As shown in Table 1, the average time consumption of the algorithm of the present invention is reduced by about 80.8% compared with the fine-resolution mesh method only, and the number of expanded nodes is much lower than that of the fine-resolution mesh method only, resulting in a significant improvement in computational efficiency. The average rotation angle is reduced by 52.3% compared to the fine-resolution mesh method alone, and the path smoothness is significantly improved; The average spatial coverage reached 6.52%, which is a significant improvement compared to the coarse-resolution grid method alone, resulting in higher spatial utilization. The average path length is similar to that of the comparison algorithm, ensuring the optimality of the path.

[0077] Furthermore, the path generated by the algorithm of this invention can flexibly fit the contour of the obstacle distribution, effectively reducing unnecessary path detours while avoiding collision risks, fully demonstrating its adaptability and superiority in complex obstacle environments.

[0078] The method proposed in this invention significantly compresses the global search space by downsampling with a coarse-resolution grid and refines the grid only in local areas containing obstacles. This avoids the high computational complexity of planning with a single fine-resolution grid and significantly reduces the average time compared to the method with only a fine-resolution grid, thereby achieving a significant improvement in computational efficiency.

[0079] By combining local fine-resolution mesh refinement with path post-processing, the number of inflection points on the path polygon is effectively reduced, the average turning angle is significantly lower than that of the fine-resolution mesh method alone, and the path smoothness is optimized; the cost of UAV attitude adjustment is reduced and flight adaptability is improved.

[0080] The obstacle projection expansion mechanism eliminates collision detection vulnerabilities caused by discretization errors, the corner constraint mechanism avoids the risk of diagonal crossing collisions, and the generated path maintains a reasonable and safe gap with the obstacle to ensure flight safety.

[0081] It should be noted that the above figures are merely illustrative of the processes included in the method according to exemplary embodiments of the present invention, and are not intended to be limiting. It is readily understood that the processes shown in the above figures do not indicate or limit the temporal order of these processes. Furthermore, it is readily understood that these processes may, for example, be executed synchronously or asynchronously in multiple modules.

[0082] Furthermore, this example implementation also provides a hierarchical search 3D airspace path planning system based on an adaptive resolution grid, which can be applied to UAVs for path planning. (Reference) Figure 7 As shown, system 70 specifically includes: The multi-resolution mesh model creation module 701 is used to construct a corresponding multi-resolution mesh model based on a 3D raster map; wherein, the multi-resolution mesh model includes: coarse resolution mesh and fine resolution mesh; The coarse path sequence calculation module 702 is used to perform global path planning on the coarse resolution grid based on the start and end point information corresponding to the current path planning task, determine the coarse path of each coarse grid node, and construct the coarse path sequence. The local window calculation module 703 is used to perform non-overlapping segmentation of the coarse path sequence; and to plan local windows that adapt to the fine resolution grid, and to determine the window entry and exit points corresponding to each local window; The local fine-grained sub-path calculation module 704 is used to identify obstacles in each local window based on a fine-resolution grid, and to call the corresponding path generation strategy based on the obstacle identification results of the local window to determine the local fine-grained sub-path corresponding to each local window. The path planning generation module 705 is used to sequentially splice the path information of each local window to obtain the path planning result; wherein, the path information of the local window includes local refined sub-paths or coarse resolution paths.

[0083] The functional implementation of each module in System 70 has been explained in detail in the corresponding method implementations, and will not be repeated here.

[0084] It should be noted that although several modules or units of the device for performing actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to embodiments of the present invention, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0085] This example implementation also provides a drone, including a drone body; and an electronic device disposed on the drone body; the electronic device may include at least a processor and a memory; wherein the memory is used to store executable instructions of the processor; the processor is configured to perform the hierarchical search three-dimensional airspace path planning based on adaptive resolution grid described in the above embodiments by executing the executable instructions.

[0086] The electronic device may include a Central Processing Unit (CPU), which performs various appropriate actions and processes based on programs stored in Read-Only Memory (ROM) or loaded from storage into Random Access Memory (RAM). RAM also stores various programs and data required for system operation. The CPU, ROM, and RAM are interconnected via a bus. Furthermore, the electronic device may include input / output (I / O) interfaces also connected to the bus.

[0087] In addition, the electronic device may include a communication section connected to a bus. The communication section performs communication processing via a satellite network. Drives are also connected to the I / O interface as needed. Removable media, such as disks, optical disks, magneto-optical disks, semiconductor memories, etc., are installed on the drive as needed so that computer programs read from them can be installed into the storage section as required.

[0088] In particular, according to embodiments of the present invention, the processes described below with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a storage medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication component, and / or installed from a removable medium. When the computer program is executed by a central processing unit (CPU), it performs various functions defined in the system of this application.

[0089] It should be noted that the storage medium shown in the embodiments of the present invention can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable compact disc read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In the present invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, wherein computer-readable program code is carried. Such transmitted data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The computer-readable signal medium can also be any storage medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the storage medium can be transmitted using any suitable medium, including but not limited to wireless, wired, etc., or any suitable combination thereof.

[0090] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0091] The units described in the embodiments of the present invention can be implemented in software or hardware, and the described units can also be located in a processor. The names of these units do not necessarily limit the specific unit itself.

[0092] It should be noted that, as another aspect, this application also provides a storage medium, which may be included in an electronic device or may exist independently without being assembled into the electronic device. The aforementioned storage medium carries one or more programs, which, when executed by an electronic device, cause the electronic device to perform the methods described in the following embodiments. For example, the electronic device may perform... Figure 1 The steps of the method shown.

[0093] In one embodiment, this application provides a computer program product including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0094] Furthermore, the above figures are merely illustrative of the processes included in the method according to exemplary embodiments of the present invention, and are not intended to be limiting. It is readily understood that the processes shown in the above figures do not indicate or limit the temporal order of these processes. Additionally, it is readily understood that these processes may be executed synchronously or asynchronously, for example, in multiple modules.

[0095] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention herein. This application is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the claims.

[0096] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is defined only by the appended claims.

Claims

1. A hierarchical search method for three-dimensional spatial path planning based on adaptive resolution grids, characterized in that, The method includes: A multi-resolution mesh model is constructed based on a 3D raster map; the multi-resolution mesh model includes: coarse-resolution mesh and fine-resolution mesh; Based on the start and end point information corresponding to the current path planning task, global path planning is performed on the coarse resolution grid to determine the coarse path of each coarse grid node and construct the coarse path sequence. The coarse path sequence is segmented into non-overlapping segments; local windows adapted to fine-resolution grids are planned, and the entry and exit points of each local window are determined. Obstacles are identified in each local window based on a fine-resolution mesh, and the corresponding path generation strategy is called based on the obstacle identification results of the local window to determine the local refined sub-path corresponding to each local window. The path information of each local window is sequentially spliced ​​together to obtain the path planning result; the path information of the local window includes local fine-grained sub-paths or coarse-resolution paths.

2. The method according to claim 1, characterized in that, Constructing a corresponding multi-resolution grid model based on a 3D raster map, including: The 3D raster map is downsampled in the horizontal direction based on a preset downsampling factor, and a coarse-resolution grid is constructed based on the sampling results. The coarse-resolution grid is downsampled horizontally based on a preset downsampling factor, and a fine-resolution grid is constructed based on the sampling results; and The projection of the obstacle onto the horizontal plane is dilated based on a preset size to obtain the obstacle mask under coarse resolution and fine resolution grids; The downsampling factor is configured based on the current terrain features corresponding to the three-dimensional raster map.

3. The method according to claim 1, characterized in that, Based on the start and end point information corresponding to the current path planning task, global path planning is performed on the coarse-resolution grid to determine the coarse path sequence, including: Based on the coarse-resolution grid occupancy mask, the improved A is called. The method solves the globally connected path between the start and end points and outputs a coarse path sequence; among them, the improved A The methods include: f(g m )=g (g m )+α・h (g m ) Where g(gm) is the cumulative Euclidean distance cost from the starting point to the current coarse grid gm, h(gm) is the Euclidean distance heuristic cost from the current coarse grid gm to the ending point, and α is the heuristic weight.

4. The method according to claim 1, characterized in that, The coarse path sequence is segmented into non-overlapping segments; local windows adapted to the fine-resolution mesh are planned, and the entry and exit points of each local window are determined, including: Configure the size of the local window based on the coarse grid node data in the coarse path sequence; Based on the size of the local window, the coarse path sequence is segmented into non-overlapping segments to obtain multiple groups of non-overlapping coarse path nodes. For each coarse path node, configure its first minimum bounding rectangle on the horizontal plane in the coarse resolution grid; and map the first minimum bounding rectangle to the fine resolution grid to determine the second minimum bounding rectangle on the fine resolution grid, so as to obtain the pixel index range of the local window based on the fine resolution grid; Based on the pixel index range of the local window using a fine-resolution grid, the preset priority rules are invoked to configure the entry and exit points of each local window.

5. The method according to claim 4, characterized in that, Configure the entry and exit points of each local window by invoking preset priority rules, including: The priority rules for selecting entry points include the following from highest to lowest priority: the starting point corresponding to the current path planning task, the exit point of the previous local window, the intersection of the skeleton path edge and the local window boundary, and the coarse resolution grid node. The priority rules for selecting exit points include the following from highest to lowest priority: the endpoint corresponding to the current path planning task, the entry point of the next local window, the intersection of the skeleton path edge and the local window boundary, and the coarse-resolution grid node.

6. The method according to claim 1, characterized in that, Obstacles are identified in each local window based on a fine-resolution mesh. Then, based on the obstacle identification results of each local window, the corresponding path generation strategy is invoked to determine the local refined sub-path for each local window, including: Identify whether there are any occupied grid cells within the current local window; and if it is determined that there are occupied grid cells, configure the current local window as a window containing obstacles; For windows containing obstacles, configure a fine-mesh search domain with the entry and exit points as the start and end points; and call the improved A function on the fine-mesh search domain. The method performs a global connected path search on the start and end points, generates continuous path segments based on a fine-resolution grid, and configures them as locally refined sub-paths. Alternatively, for an accessible window, configure the coarse path node corresponding to the local window as a continuous path segment corresponding to the accessible window, and configure it as a local refined sub-path.

7. The method according to claim 6, characterized in that, For windows containing obstacles, configure a fine-mesh search domain with the entry and exit points as the starting and ending points, including: Based on the coordinate information of the entry and exit points under fine-resolution mesh, the minimum bounding rectangle of the horizontal plane is generated; wherein, the minimum bounding rectangle is determined based on the extreme values ​​of the coordinates of the entry and exit points; The minimum bounding rectangle of the horizontal plane is superimposed with a height range to form an axis-aligned bounding box, which is then configured as the base search region based on a fine-resolution grid.

8. The method according to claim 7, characterized in that, The method further includes: The basic search region is uniformly expanded outward in the horizontal direction based on the expansion parameter S, and the expanded region is configured as the fine mesh search domain.

9. The method according to claim 1, characterized in that, The method further includes: When the aircraft moves based on the path planning results, identify the diagonal grid of the current grid or the adjacent grids of each axis of the spatial diagonal; If any adjacent grid along an axis contains an obstacle, movement to that diagonal grid or spatial diagonal is prohibited.

10. A hierarchical search three-dimensional spatial path planning system based on adaptive resolution grids, characterized in that, The system includes: The multi-resolution mesh model creation module is used to construct corresponding multi-resolution mesh models based on 3D raster maps; the multi-resolution mesh models include: coarse-resolution meshes and fine-resolution meshes; The coarse path sequence calculation module is used to perform global path planning on the coarse resolution grid based on the start and end point information corresponding to the current path planning task, determine the coarse path of each coarse grid node, and construct the coarse path sequence. The local window calculation module is used to perform non-overlapping segmentation of the coarse path sequence; and to plan local windows that adapt to the fine resolution grid, and determine the window entry and exit points corresponding to each local window; The local fine-grained sub-path calculation module is used to identify obstacles in each local window based on a fine-resolution grid, and to call the corresponding path generation strategy based on the obstacle identification results of the local window to determine the local fine-grained sub-path corresponding to each local window. The path planning generation module is used to sequentially splice the path information of each local window to obtain the path planning result; the path information of the local window includes local fine-grained sub-paths or coarse-resolution paths.

Images