An unmanned aerial vehicle patch investigation route planning method and device

Abstract

The application provides a kind of unmanned aerial vehicle map spot investigation route planning method and device, wherein the method comprises: obtaining the vector polygon boundary data of the region to be investigated, and constructing target surface and planning surface with different task semantics and geometric characteristics;According to the ground coverage size of a single photo calculated by camera parameters and ground sampling distance, classify the map spot;According to the determined map spot type, adopt the matching route generation strategy to generate the classified route;Based on the classified route, generate the discrete flight point set of unmanned aerial vehicle, and decouple the action arrangement of spatial position and evidence attitude.Through the application, the flight and photographing time of unmanned aerial vehicle in non-target area is reduced, the coverage efficiency is improved, and the problem of poor efficiency and accuracy of map spot investigation in the prior art is solved.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) mission planning technology, and in particular to a method and apparatus for planning flight routes for UAV map surveys. Background Technology

[0002] In existing UAV-based aerial survey mission planning, the common approach is to obtain the bounding rectangle (e.g., the minimum bounding rectangle or bounding box) for the map patch, generate equally spaced parallel scan lines inside the rectangle, and fly back and forth in a reciprocating scanning manner, while triggering photography at equal time or equal distance intervals.

[0003] The above methods are applicable to regular, approximately convex, and wide planar features, but in actual planar feature survey operations such as land and resources satellite imagery enforcement and land change surveys, they face the following significant problems: First, the complex shapes of the patches (such as winding rivers, L-shaped, cross-shaped, hook-shaped, etc., long strips / corridor-shaped patches) and the inclusion of a large number of non-target areas in the bounding rectangles lead to low coverage efficiency. In addition, the frequent back-and-forth turns result in a large number of turns and high maneuvering costs.

[0004] Secondly, the core purpose of the land parcel survey is not only to obtain orthophotos, but also to clearly identify the boundaries of the land parcels, the sides of illegal structures, or the condition of land features. Traditional reciprocating scanning often uses vertical downward viewing or random yaw angles at the boundaries, which easily leads to the omission of side features of edge features (such as building facades), and when the scan line direction is mismatched with the boundary, edge omissions and redundancy are likely to coexist.

[0005] There is currently no effective solution to the problem of poor efficiency and accuracy in existing related technologies for mapping. Summary of the Invention

[0006] This invention provides a method and apparatus for planning flight routes for UAV map surveys, which addresses the shortcomings of poor efficiency and accuracy in existing related technologies for map surveys.

[0007] In a first aspect, the present invention provides a method for planning flight routes for UAV map surveys, comprising: Obtain vector polygon boundary data of the area to be investigated, and construct target surfaces and planning surfaces with different task semantics and geometric features; The ground coverage size of a single photo is calculated based on camera parameters and ground sampling distance, and the patches are classified. Based on the determined map patch type, a matching route generation strategy is adopted to generate categorized routes; Based on the classified routes, a set of discrete waypoints for the UAV is generated, and action choreography is performed to decouple spatial position from evidence collection attitude.

[0008] According to the present invention, a method for planning flight routes for UAV patch surveys involves acquiring vector polygon boundary data of the area to be surveyed and constructing target surfaces and planning surfaces with different task semantics and geometric features, including: Obtain vector polygon boundary data in the area to be investigated; Transform the vector polygon boundary data from the geographic coordinate system to the working projection coordinate system; In the working projection coordinate system, a planning surface for path generation and a target surface for coverage acceptance are constructed simultaneously.

[0009] According to the present invention, a method for planning flight routes for UAV patch surveys simultaneously constructs a planning surface for path generation and a target surface for coverage acceptance in the working projection coordinate system, including: Perform geometric repair or normalization on the input polygon, process internal hole features, and perform buffered expansion according to the outward expansion margin; The processing of internal hole features includes removing the internal hole features of polygons and retaining only the outermost envelope boundary, so that the area enclosed by the internal holes is incorporated into the planning surface and the target surface; The target surface is used for calculating the coverage of UAV aerial surveys and for final result acceptance, while the planning surface is used to guide the generation of the underlying UAV flight path.

[0010] According to the present invention, a method for planning flight routes for UAV patch surveys calculates the ground coverage size of a single photograph based on camera parameters and ground sampling distance, and classifies the patches, including: The ground coverage width and height of a single photo are calculated based on the camera parameters and the ground sampling distance, and the flight altitude, strip spacing and heading sampling spacing are determined accordingly. A hierarchical filtering strategy, ranging from low to high computational complexity, is used to classify the patches; the types of patches include small patches, strip or corridor patches, and regular patches.

[0011] According to the present invention, a method for planning flight routes for UAV map patch surveys employs a hierarchical filtering strategy from low to high computational complexity to classify map patches, including: If the patch satisfies the criterion of a narrow rectangle based on the minimum circumscribed rotating rectangle, then it is determined whether the patch is a strip or a corridor-type patch. If the patch does not satisfy the narrow rectangle criterion based on the minimum bounding rotation rectangle, but satisfies the initial screening and interception strategy based on the morphological equivalent width, then it is determined whether the patch is a regular patch. If the patch does not meet the initial screening interception strategy based on morphological equivalent width, but meets the fine criteria of corridor skeleton, then the patch is determined to be a strip or corridor-type patch.

[0012] According to the present invention, a method for planning flight routes for UAV patch surveys generates classified flight routes based on the determined patch type using a matching flight route generation strategy, including: When the patch is determined to be a small patch, it supports generating a basic waypoint sequence from a single orthophoto overlay; When the patch is determined to be a strip or corridor-type patch, the skeleton network is extracted inside the planning surface and a node graph model is constructed. The end-to-end trunk path is obtained through the trunk path search algorithm, and the centerline waypoint sequence is obtained by resampling the end-to-end trunk path. When the patch is determined to be a regular patch, a basic waypoint sequence is generated using a reciprocating scanning method.

[0013] According to the present invention, a method for planning flight routes for UAV patch surveys includes extracting a skeleton network within the planning surface and constructing a node graph model, comprising: The boundary points are sampled at arc length steps to obtain a set of boundary points, and a Voronoi diagram of the boundary points is constructed. The edges of the Voronoi diagram are taken and clipped into the planning plane to obtain a set of candidate skeleton line segments. Non-main line segments with a length less than a preset threshold and small redundant branches that are too close to the boundary are removed to obtain the skeleton line network. The skeleton network is sampled at equal intervals to obtain a set of nodes, and spatially nearest nodes are fused based on a preset distance threshold. Connect nodes with undirected edges according to their adjacent radii, and use Euclidean distance as the edge weight; Each node is limited to retaining a maximum of a preset number of adjacent edges that are closest to the node, and the largest connected component is selected as the main graph of the node graph model.

[0014] According to the present invention, a method for planning flight routes for UAV map surveys includes obtaining an end-to-end trunk path through a trunk path search algorithm and resampling the end-to-end trunk path to obtain a centerline waypoint sequence, comprising: In the node graph model, a starting point is selected, and the first shortest path search is performed to obtain the farthest first endpoint. Using the first endpoint as the source point, perform a second shortest path search to obtain the second endpoint that is furthest away; The path between the first endpoint and the second endpoint is reconstructed based on the predecessor relationship of the second shortest path search as the end-to-end backbone path; The cumulative arc length is calculated for the end-to-end trunk path, and an arc length sequence is generated according to a preset sampling interval. Resampled waypoints are obtained by linear interpolation of the arc length on adjacent broken line segments, and a centerline waypoint sequence is generated.

[0015] According to the present invention, a method for planning flight routes for UAV map surveys generates a set of discrete waypoints for the UAV based on the classified flight routes, and performs action choreography that decouples spatial position from evidence-gathering attitude, including: Assign waypoint role identifiers to waypoints in the discrete waypoint set and configure arrival action sequences corresponding to the waypoint role identifiers; the arrival action sequences include at least tilt observation and forward / backward tilt observation attitude control for map boundaries and specific waypoint roles, in order to obtain the three-dimensional elevation features of the target features; The output includes the three-dimensional spatial location and the sequence of actions to the point, and controls the UAV to perform the three-dimensional map evidence collection task of the area to be investigated.

[0016] Secondly, the present invention also provides a UAV map survey route planning device, comprising: The construction module is used to acquire vector polygon boundary data of the area to be investigated, and to construct target surfaces and planning surfaces with different task semantics and geometric features; The classification module is used to calculate the ground coverage size of a single photo based on camera parameters and ground sampling distance, and to classify the patches. The decision-making module generates categorized routes by adopting a matching route generation strategy based on the determined map patch type. The planning module is used to generate a set of discrete waypoints for the UAV based on the classified routes, and to perform motion orchestration that decouples spatial position from evidence collection attitude.

[0017] Thirdly, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the UAV map survey route planning method as described in the first aspect above.

[0018] Fourthly, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the UAV map survey route planning method as described in the first aspect above.

[0019] Fifthly, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the UAV map survey route planning method as described in the first aspect above.

[0020] Compared with the prior art, the present invention has the following beneficial effects: 1. The UAV map survey route planning method provided by the invention reduces flight and photography in non-target areas and improves coverage efficiency in long and narrow / corridor-shaped, curved and bifurcated map patches by planning the main skeleton centerline. This method avoids frequent turning back in corridor map patches by using an end-to-end main skeleton path, reduces the number of turns and attitude changes, improves flight control executability and reduces mission time.

[0021] 2. This method closely aligns with the operational needs of land parcel surveys. Through outward buffering and attitude control at entry, exit, and flight path edges, it effectively acquires the facade and overall features of illegal structures and land parcel boundaries, significantly improving the quality of evidence collection. Furthermore, by separating the target surface from the planned surface, this method improves the stability of path generation while ensuring strict alignment of the target area during coverage acceptance. Through fully discrete waypoints and action orchestration, the semantics of the survey segment and transition segment are separated, enhancing mission interpretability, front-end display consistency, and flight control protocol compatibility.

[0022] 3. This method achieves coverage acceptance and missed coverage detection through a computable ground footprint model, facilitating debugging, quality control, and automated acceptance. In multi-patch tasks, the order of patch visits can be determined and the order information can be output, improving the determinism and reproducibility of task generation and reducing flight overhead in cross-patch transition sections.

[0023] 4. This method possesses high robustness and practical engineering value. Verified through large-scale real-world scenarios (7477 uninvestigated map patches in a region with complex terrain and diverse land parcels), this solution, by combining footprint validity screening and adaptive coverage repair mechanisms, achieves a 0% omission rate in the target area when facing complex map patch morphologies such as extreme irregularities, elongated bifurcations, and high depressions. This provides robust data integrity assurance for automated map patch investigation and evidence collection. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0025] Figure 1 This is a flowchart of the UAV map survey route planning method provided by the present invention; Figure 2 This is a schematic diagram of the unmanned aerial vehicle (UAV) map survey route planning process in an embodiment of the present invention; Figure 3 This is a comparison diagram of the construction of the target surface and the planned surface set in an embodiment of the present invention; Figure 4This is a flowchart illustrating the process of generating routes for strip or corridor-type patches in an embodiment of the present invention; Figure 5 This is a schematic diagram of adaptive point filling for bifurcation and missing coverage areas in an embodiment of the present invention; Figure 6 This is a schematic diagram of the discrete waypoint roles and actions in an embodiment of the present invention; Figure 7 This is a schematic diagram of the footprint projection using the ray method in an embodiment of the present invention; Figure 8 This is a schematic diagram illustrating the connection between multi-spot task determination and global waypoints in an embodiment of the present invention; Figure 9 This is a schematic diagram illustrating coverage verification in an embodiment of the present invention; Figure 10 This is a schematic diagram of the actual map strip route generation result in an embodiment of the present invention; Figure 11 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0027] This invention provides a method for planning flight routes in UAV map surveys. Figure 1 This is a flowchart of the UAV map survey route planning method provided by the present invention, such as... Figure 1 As shown, the method includes the following steps: Step S101: Obtain vector polygon boundary data of the area to be investigated, and construct target surfaces and planning surfaces with different task semantics and geometric features; the vector polygon boundary data includes multiple polygonal patches; Step S102: Calculate the ground coverage size of a single photo based on camera parameters and ground sampling distance (GSD), and classify the patches; Step S103: Based on the determined map patch type, a matching route generation strategy is adopted to generate classified routes; Step S104: Generate a set of discrete waypoints for the UAV based on the classified flight path, and perform motion choreography that decouples the spatial position from the evidence collection attitude.

[0028] In this method, firstly, the vector polygon boundary data of the area to be investigated is projected onto the working coordinate system to construct the target surface and the planning surface. Then, the single-image coverage size is calculated based on camera parameters and GSD, and the patches are classified. Next, based on the patch type, classified flight paths are generated according to the corresponding flight path generation strategy. Finally, a set of discrete waypoints for the UAV is generated based on the classified flight paths, and motion choreography is performed to decouple spatial position from evidence-gathering attitude. Through the above process, corresponding flight paths are planned for different types of patches, reducing the UAV's flight and photography time in non-target areas, improving coverage efficiency, and solving the problems of poor patch survey efficiency and accuracy in existing related technologies.

[0029] Figure 2 This is a schematic diagram of the unmanned aerial vehicle (UAV) map survey route planning process in an embodiment of the present invention, such as... Figure 2 As shown below, the above method will be explained in detail: In some embodiments, step S101, acquiring vector polygon boundary data of the area to be investigated, and constructing target surfaces and planning surfaces with different task semantics and geometric features, includes: acquiring vector polygon boundary data of the area to be investigated; converting the vector polygon boundary data from the geographic coordinate system to the working projection coordinate system; and simultaneously constructing a planning surface for path generation and a target surface for coverage acceptance in the working projection coordinate system.

[0030] In this embodiment, the vector polygon boundary data includes polygons or multi-polygons in GeoJSON format, as well as task parameters such as camera parameters, GSD, overlap rate, outer margin, and transition height. The coordinates of the vector polygon boundary data are ordered pairs of longitude and latitude. The vector polygon boundary data is transformed from the geographic coordinate system to a projected coordinate system in meters to ensure that the dimensions of geometric operations such as distance, area, and buffer remain consistent. The geographic coordinate system can be the WGS84 coordinate system, and the projected coordinate system can be the UTM partition corresponding to the takeoff point. Projecting the input polygon to the working coordinate system facilitates calculations of area, perimeter, buffer, and distance, and the final output is then transformed back to the geographic coordinate system.

[0031] For each map patch, a corresponding discrete waypoint subsequence is generated, and the discrete waypoint subsequences are concatenated into a global discrete waypoint sequence according to the map patch access order. The map patch access order is determined according to the received sequence identifier sequence, or if no sequence identifier sequence is received, it is determined by using a local optimization strategy based on the shortest spatial distance priority and combining it with the centroid of the representative point of the map patch, starting from the takeoff point.

[0032] Simultaneously construct a planning surface for path generation and a target surface for coverage acceptance in the working projection coordinate system, including: geometric repair or normalization of the input polygon, processing of internal hole features, and buffering expansion according to the outer expansion margin; processing of internal hole features includes removing the internal hole features of the polygon and retaining only the outermost envelope boundary, so that the area enclosed by the internal holes is incorporated into the planning surface and the target surface.

[0033] Figure 3 This is a comparison diagram of the target surface and the planned surface set construction in an embodiment of the present invention, such as... Figure 3 As shown, the target surface is defined as the absolute set benchmark used for UAV aerial survey coverage calculation and final result acceptance. Its construction logic is as follows: geometric repair and normalization are performed on the input polygon, all internal hole features are removed (only the outermost envelope boundary is retained), and polygon buffering is performed outward according to the preset outer margin parameters. The outer buffering and target surface construction aim to ensure that in land survey operations, the edges of ground features and the surrounding reference environment are completely within the effective field of view, meeting strict boundary verification standards.

[0034] The planning surface is defined as the geometric and physical carrier used to guide the generation of the UAV's underlying flight path. Its initial construction logic is consistent with that of the target surface (i.e., it also undergoes hole removal and outward buffering). Furthermore, when the subsequent process determines that the patch type is a non-strip or corridor type patch, a convex hull operation is forcibly performed on the surface to generate the smallest convex polygon containing all vertices of the surface as the final planning surface. This eliminates the fragmentation problem of the scanning flight path caused by severely concave boundaries at the algorithm level, and greatly improves the stability of the UAV's continuous flight.

[0035] In some embodiments, step S102, which calculates the ground coverage size of a single photograph based on camera parameters and ground sampling distance, and classifies the patches, includes: calculating the ground coverage width and height of a single photograph based on camera parameters and ground sampling distance, and determining the flight altitude, strip spacing, and heading sampling spacing accordingly; and using a hierarchical filtering strategy from low to high computational complexity to classify the patches in the area to be investigated. The types of patches include small patches, strip or corridor-type patches, and regular patches.

[0036] For example, given GSD, focal length f Sensor size With image pixel size From width to width w With Gao Xiang h Derivation of flight altitude:

[0037] in, This represents the relative flight altitude relative to the ground, derived from the image width. Indicates the image pixel width. Indicates focal length. Indicates the physical width of the sensor. This represents the relative flight altitude relative to the ground, derived from the image altitude. Indicates the image pixel height. This represents the physical height of the sensor. And take:

[0038] in, This indicates the final determined relative flight altitude with respect to the ground. Single orthophoto ground coverage size. for:

[0039] in, This indicates the width of the ground coverage in a single photograph. Indicates the ground cover height of a single photograph. Strip spacing. Sampling interval with heading They can be respectively:

[0040] in, Indicates the strip spacing. Indicates the heading sampling interval. This represents the heading overlap parameter, which ranges from (0, 1). In one embodiment, the heading overlap parameter... A value of 0.15 is available, and configuration is allowed based on task requirements.

[0041] A hierarchical filtering strategy, ranging from low to high computational complexity, is employed to classify map features. This includes: if a feature satisfies the criterion of a narrow rectangle based on the minimum bounding rectangle of rotation, it is determined whether the feature is a strip or corridor-type feature; if a feature does not satisfy the criterion of a narrow rectangle based on the minimum bounding rectangle of rotation, but satisfies the initial screening and interception strategy based on morphological equivalent width, it is determined whether the feature is a regular feature; if a feature also does not satisfy the initial screening and interception strategy based on morphological equivalent width, but satisfies the fine-grained corridor skeleton criterion, it is determined whether the feature is a strip or corridor-type feature.

[0042] For example, when the short side length of the smallest bounding rectangle of the target surface is not greater than the short side length of a single cover, and its long side length is not greater than the long side length of a single cover, the patch is determined to be a small patch to support single orthophoto coverage.

[0043] To balance the accuracy of patch morphology recognition with the computational overhead of batch processing, a hierarchical filtering strategy is adopted for the determination of strip or corridor-type patches, and the following criteria are applied in order of increasing computational complexity: First layer: Lightweight criterion based on minimum bounding rectangle (slender rectangle criterion): This level aims to quickly filter out regular, straight stripes with extremely low geometric computational overhead. A threshold for the degree of thinness is set as follows: (It requires a value greater than 1, for example, 5.0); the upper limit of the effective coverage width of a single flight strip is denoted as... ,in ( (For example, a safety redundancy factor of 0.9). Calculate the minimum bounding rectangle of the target surface. When the aspect ratio of this minimum bounding rectangle exceeds the threshold... And its shorter side length does not exceed the upper limit of the effective coverage width. If the pattern is not clear, it is directly identified as a strip or corridor-shaped patch and subsequent detection is skipped.

[0044] Second layer: Preliminary screening based on morphological equivalent width (short-circuit evaluation interception): When the polygon exhibits curved (e.g., "U" or "S" shape) or multi-directional branching, the minimum bounding rectangle will leave a large area of ​​blank space, causing the first-level criterion to fail. In this case, to avoid blindly introducing time-consuming topological skeleton extraction operations, a morphological equivalent width is introduced for low-cost initial screening: based on the area of ​​the polygon. A With perimeter P Construct an equivalent width estimate :

[0045] in, This represents the estimated equivalent width. A Represents the area of ​​a polygon. P Represents the perimeter of the polygon. Determines the estimated equivalent width. Does it exceed the preset tolerance multiple (e.g.) If the width exceeds the limit, it indicates that the main body of the patch is too wide and it is directly judged as a regular patch; if the width does not exceed the limit, it proceeds to the third layer of core topology detection.

[0046] Third layer: Detailed criteria for corridor framework (addressing irregular bends and forks): For the patches that pass the second initial screening, high-precision skeleton mesh extraction is performed, and the final judgment is made in combination with the skeleton topological features: Extraction and Calculation: Extract the skeleton mesh inside the polygon and calculate the total length of the skeleton. Simultaneously, dense sampling is performed on the skeleton, and twice the distance from each sampling point to the boundary of the patch is calculated as the local width estimate. The preset quantile of its width distribution (e.g., the 90th quantile, denoted as ) is then taken. () serves as the upper bound of the actual width of the patch.

[0047] Corridor Ratio Test: Constructing a Corridor Ratio Index :

[0048] in, Indicates the corridor ratio index, Indicates the total length of the skeleton. Represents the area of ​​a polygon.

[0049] When the corridor ratio index It is greater than the preset shape threshold, and its true width upper bound is greater than the upper bound. Not exceeding the upper limit of the effective coverage width When the condition is met, it is determined to be a strip or corridor-type feature. Feature other than small features and strip or corridor-type features are determined to be regular features.

[0050] Based on this, in step S103, according to the determined type of map patch, a matching route generation strategy is adopted to generate classified routes, including: when the map patch is determined to be a small map patch, a single orthophoto cover is used to generate a basic waypoint sequence; when the map patch is determined to be a strip or corridor type map patch, a skeleton network is extracted inside the planning surface and a node graph model is constructed, an end-to-end trunk path is obtained through a trunk path search algorithm, and the end-to-end trunk path is resampled to obtain a centerline waypoint sequence; when the map patch is determined to be a regular map patch, a reciprocating scanning method is used to generate a basic waypoint sequence.

[0051] The process of extracting a skeleton network and constructing a node graph model within the planning plane includes: sampling the boundary points of the planning plane at arc length steps to obtain a set of boundary points, and constructing a Voronoi diagram of the boundary points; taking the edges of the Voronoi diagram and trimming them to the planning plane to obtain a set of candidate skeleton segments, and removing non-mainline segments with lengths less than a preset threshold and small redundant branches too close to the boundary to obtain the skeleton network; sampling the skeleton network at equal intervals to obtain a set of nodes, and performing fusion processing on spatially nearest nodes based on a preset distance threshold; connecting undirected edges between nodes according to their adjacency radius, using Euclidean distance as the edge weight; limiting each node to retain at most a preset number of adjacent edges closest to the node, and selecting the largest connected component as the main graph of the node graph model. The adjacency radius is dynamically calculated based on the distance between the centerline waypoints, and is positively correlated with the distance along the line. The adjacency radius is truncated based on preset upper and lower distance limits; the preset number is an integer greater than 0.

[0052] An end-to-end backbone path is obtained through a backbone path search algorithm, and the centerline waypoint sequence is obtained by resampling the end-to-end backbone path. This process includes: selecting a starting point in the node graph model; performing a first shortest path search to obtain the farthest first endpoint; performing a second shortest path search using the first endpoint as the source point to obtain the farthest second endpoint; reconstructing the path between the first and second endpoints based on the predecessor relationships of the second shortest path search as the end-to-end backbone path; calculating the cumulative arc length of the end-to-end backbone path; generating an arc length sequence according to a preset sampling interval; and obtaining resampled waypoints by linear interpolation of the arc length on adjacent polyline segments to generate the centerline waypoint sequence. The preset sampling interval is determined based on the ground coverage size and overlap rate of a single photograph.

[0053] For example, when a patch type is determined to be a small patch, it indicates that the ground coverage of a single image is sufficient to encompass the entire patch. Therefore, without complex scan route reversals or skeleton extraction, the following single-image orthophoto cover generation steps can be performed: 1. Waypoint 2D Coordinate Determination: Calculate the geometric feature points of the patch as the 2D plane coordinates of the base waypoint. In one embodiment, the geometric feature points are taken as the center point of the smallest bounding rectangle of the target surface. Using the center point of the smallest bounding rectangle of the bounding rectangle instead of the centroid of the polygon ensures that the single rectangular coverage footprint generated according to the camera's aspect ratio can most uniformly and symmetrically encompass the target surface boundary in all directions, thereby minimizing the risk of local omissions caused by the extremely irregular shape of the patch.

[0054] 2. Waypoint Altitude Assignment: Assign orthorectified coverage flight altitudes to the base waypoints. These orthorectified coverage flight altitudes are strictly calculated based on preset parameters such as ground sampling distance, camera focal length, and sensor size to meet the stringent ground resolution standards required for the underlying map patch survey operation.

[0055] 3. Basic Waypoint Sequence Output: The three-dimensional spatial points with two-dimensional planar coordinates and orthophoto altitude information are output as a single basic waypoint sequence for the small patch. This basic waypoint sequence only represents the spatial position reference of the flight platform. In the subsequent "fully discrete waypoints and action orchestration" step, it will be further assigned specific waypoint role identifiers and attached with corresponding arrival action sequences (such as triggering a single orthophoto, or optionally inserting a central panoramic action).

[0056] Figure 4 This is a flowchart of the strip or corridor-type map route generation process in an embodiment of the present invention, such as... Figure 4 As shown, when the map feature is a strip or corridor-shaped map feature, the following steps are performed: 1. Skeleton mesh extraction and trimming: The outer boundary of the planning surface is sampled according to the arc length step to obtain the boundary point set. The Voronoi diagram of the boundary point set is constructed. The Voronoi edges are taken and clipped to the interior of the planning surface to obtain the candidate skeleton line segment set. Non-main line segments with a length less than a preset threshold and small redundant branches that are too close to the boundary are further removed to obtain the skeleton line network, such as a line network geometric object composed of multiple line segments.

[0057] 2. Point graph construction: The skeleton network is sampled at equal intervals according to the step size to obtain a set of nodes. Nearest neighbor nodes are then fused based on a preset distance threshold. Undirected edges are connected between nodes according to their adjacency radius, and Euclidean distance is used as the edge weight to form a weighted undirected graph. To control the sparsity and determinism of the graph, only the nearest N adjacent edges are retained for each node. The connected component with the most nodes or the longest skeleton length is selected as the main graph.

[0058] 3. End-to-end backbone extraction: A deterministic starting point is selected in the main graph, and a first shortest path search (using Dijkstra's algorithm) is performed to obtain the farthest first endpoint. Using the first endpoint as the source point, a second shortest path search is performed to obtain the farthest second endpoint. Based on the predecessor relationships from the second shortest path search, the path from the first endpoint to the second endpoint is reconstructed as the end-to-end backbone path. This step aims to prioritize continuous flight in the main extension direction of the corridor, but in multi-branch structures, non-backbone lateral branches are temporarily discarded.

[0059] 4. Arc length resampling generates centerline waypoints: Connect the main paths into a broken line, and resample the broken line at equal intervals according to the arc length. The sampling interval can be the waypoint spacing s along the line or its equivalent form to obtain the centerline waypoint sequence. Assign a uniform orthogonal coverage flight altitude to each waypoint to initially form the basic waypoint sequence. The waypoint spacing s along the line can be the same as the sampling interval along the heading. The same calculation method.

[0060] 5. Adaptive point filling for bifurcation and missed coverage areas: Figure 5 This is a schematic diagram of adaptive point filling for bifurcation and missing coverage areas in an embodiment of the present invention, as shown below. Figure 5 As shown, to address the physical omissions caused by lateral long branches temporarily discarded during end-to-end backbone extraction, an iterative repair process based on the criteria of minimizing the local minimum bounding rectangle and the newly added mileage is designed: 5.1 Valid Point Filtering and Coverage Calculation: Obtain the initial centerline waypoint sequence, and calculate the orthorectified footprint polygon for each waypoint based on the ground coverage model of a single photograph. After removing invalid waypoints whose orthorectified footprints do not intersect the target surface, the union of the orthorectified footprints of the remaining waypoints is calculated to obtain the currently covered area.

[0061] 5.2 Gap Extraction and Priority Selection: Perform a set difference operation on the target area and the currently covered area to obtain the set of missing gaps. If the total area of ​​the missing gaps is equal to 0, skip the repair process; otherwise, extract the gap block with the largest area as the priority repair target for the current round.

[0062] 5.3. Generation of Adaptive Gap-Fitting Sequences: Calculate the minimum circumscribed rotation rectangle (MRR) of the priority repair target, and extract the long side direction of the MRR as the main axis direction for reshooting to adapt to the extension shape of the gap. The camera moves laterally along a direction perpendicular to the main axis of the reshoot, and generates multiple parallel reshoot trajectories based on the effective footprint width and the lateral overlap rate. Sampling is performed on each reshoot trajectory based on footprint height and longitudinal overlap rate. The generated reshoot waypoints are connected in a reverse serpentine manner to output a coherent local reshoot subsequence.

[0063] 5.4 Minimum Cost Path Insertion: Traverse each adjacent node pair (i.e., candidate insertion segment) in the currently retained main waypoint sequence. Calculate the additional flight mileage generated when the local supplementary subsequence is interrupted in both "forward" and "reverse" directions and inserted into each adjacent node pair. Select the adjacent node pair with the minimum additional flight mileage and its corresponding sequence direction for insertion, so that the route can compensate for lateral branch coverage while maintaining the smoothness and flight efficiency of the original path to the greatest extent.

[0064] 5.5 Iterative Loop Closure: Mark the newly inserted supplementary waypoints as coverage repair points, return to step 5.1 to recalculate coverage and gaps, until the missed coverage gaps meet the preset tolerance, or the set upper limit for the number of repair points is reached. After the iterative loop is completed, output a complete sequence containing the backbone and supplementary waypoints, as the basic waypoint sequence for this strip or corridor-type patch.

[0065] When the patch type is a conventional patch, the survey track is generated using a reciprocating scanning method. The specific steps are as follows: 1. Determining the main scanning direction: Calculate the minimum circumscribed rectangle of the planning surface, take the direction of its longest side as the main scanning direction, and normalize the main scanning direction to [0, 180). 2. Coordinate rotation and range determination: Rotate the planning surface around its centroid or preset origin until the scanning main direction is aligned with the x-axis, and calculate the rotated planning surface; 3. Strip position generation: In the rotating coordinate system, take the range of values ​​of the rotated planning surface in the normal direction. Using the strip spacing as the step size, and taking this range as a reference, a sequence of centered strip positions is generated to both sides, thereby covering the... To avoid creating extremely short boundary stripes due to edge contact; 4. Candidate Survey Point Sampling: For each strip, in the rotating coordinate system along the scanning direction, candidate survey points are generated according to the lateral intercept range of the planned surface at the corresponding y-coordinate of the strip after rotation, and the photo is triggered in the orthogonal state. When the survey point is close to the entry point, the action of the gimbal returning to the orthogonal angle can be inserted before taking the photo. Furthermore, if it is the edge survey point of the regular patch reciprocating scanning route (i.e., the start and end points of each scanning line), the fuselage yaw angle is controlled to conduct the opposite observation along the axis of the current scanning line: Specifically, at the edge point at the beginning of the scanning line, the fuselage yaw angle is kept consistent with the current scanning flight direction (forward-looking evidence collection); at the edge point at the end of the scanning line, the fuselage yaw angle is controlled to reverse to deviate from the current flight direction (backward-looking evidence collection), and combined with the gimbal downward movement, the elevation features of the plot boundary along the scanning line direction are enhanced.

[0066] 5. Footprint Validity Screening (Removal of Invalid Points): For each candidate survey point, construct its corresponding orthophoto ground cover rectangular footprint; perform an intersection test on the candidate survey point using a target surface that may contain original concave polygon features. When the rectangular footprint intersects with the target surface, the candidate survey point is retained; otherwise, if the rectangular footprint falls completely outside the target surface, it is removed. This step aims to avoid invalid photography in recessed areas or non-target areas, while allowing valid survey points to be slightly outside the boundary to ensure edge coverage.

[0067] 6. Strip sorting and reciprocating direction: Strips are generated sequentially according to their position sequence, and the reciprocating direction is set alternately for adjacent strips to form a reciprocating track; the reserved points in each strip are sorted according to their x-coordinates, and the ascending or descending order is determined according to the parity of the strips.

[0068] 7. Invalid Segment Crossing and Connecting: When the patch is a concave polygon (such as a U-shaped or ring-shaped structure), the footprint validity screening based on the target surface will cause a single continuous scan strip to be truncated into multiple discontinuous valid image segments. To maintain the continuity of the underlying flight control commands, all retained points within the same strip are sequentially connected in a pre-sorted deterministic order; for discontinuous valid image segments, straight line segments are used to cross and connect them. The crossing and connecting segments are allowed to pass through the external area of ​​the target surface, so that when the UAV crosses non-target areas, it only maintains the continuity of the flight path and the flight transition, without triggering data acquisition actions.

[0069] 8. Altitude Assignment and Sequence Output: Assign the calculated orthophoto coverage flight altitude to all the retained survey points generated in the above steps, and output the set of three-dimensional waypoints concatenated in a deterministic order as the basic waypoint sequence for this regular patch.

[0070] At this point, spatial location benchmarks have been generated for all different types of map features, providing basic input for subsequent unified arrangement of evidence collection actions.

[0071] In some embodiments, step S104 generates a discrete waypoint set for the UAV based on the classified flight path and performs action orchestration that decouples spatial position from evidence collection attitude, including: assigning waypoint role identifiers to waypoints in the discrete waypoint set and configuring arrival action sequences corresponding to the waypoint role identifiers; the arrival action sequence includes at least tilt observation and forward / reverse tilt observation attitude control for the boundary of the map patch and specific waypoint roles to obtain the three-dimensional elevation features of the target ground features; outputting the task planning result containing the three-dimensional spatial position and arrival action sequence, and controlling the UAV to perform the three-dimensional map patch evidence collection task in the area to be investigated.

[0072] In this embodiment, a discrete waypoint set is generated based on the centerline waypoint sequence or conventional patch paths; waypoint role identifiers are configured for each waypoint, and a sequence of actions is configured to reach each waypoint, including transition points, entry points, survey points, exit points, and optional central panoramic viewpoints. The action types include at least gimbal attitude adjustment, fuselage yaw setting, photo triggering, and panoramic photo triggering. Actions can be executed sequentially within the waypoints according to their action numbers; and attitude control (pointing the yaw angle inwards towards the patch) enhances the acquisition of patch edge and elevation features.

[0073] The waypoints in the discrete waypoint set include at least transition waypoints, entry waypoints, survey waypoints, and exit waypoints; the altitude of the transition waypoints is set as transition altitudes; the entry waypoints and exit waypoints are located outside the survey area determined by the polygon boundary data and a reserved distance is set along the flight direction, the reserved distance being determined based on the ground cone footprint offset of the camera at a preset tilt angle, to provide a field of view retreat space for oblique photography; the arrival action sequence includes at least one or more combinations of gimbal tilt angle adjustment action, fuselage yaw angle setting action, and photo triggering action.

[0074] The fuselage yaw angle setting and gimbal pitch angle adjustment actions in the arrival action sequence further include: for the arrival waypoint and the departure waypoint, calculating the direction vector of the waypoint relative to the interior of the area to be investigated, and setting the fuselage yaw angle to point towards the interior of the area to be investigated along the direction vector, while controlling the gimbal pitch angle to adjust to the preset tilt shooting angle; for the edge points of the regular patch reciprocating scanning route in the survey waypoint, controlling its fuselage yaw angle to conduct opposing observations along the axis direction of the current scan line (including forward and reverse directions), and combining the gimbal downward action to obtain the elevation features of the land parcel boundary along the scan line direction.

[0075] For example, such as Figure 6 As shown, Figure 6 This is a schematic diagram of the discrete waypoint roles and actions in an embodiment of the present invention. To facilitate front-end adjustment and display, this embodiment does not use continuous, equally spaced photo-taking commands, but generates fully discrete waypoints: each photo-taking intention corresponds to an independent waypoint, and a waypoint role identifier and arrival action sequence are assigned to each waypoint. Especially for land parcel survey operations, orthogonal and oblique evidence-gathering postures are designed at special waypoints to effectively obtain the facade and three-dimensional spatial features of target features such as land parcel boundaries and illegal building sides, meeting the business needs of comprehensive evidence collection.

[0076] For UAV transition altitude control: set the orthophoto coverage flight altitude and the cross-plot transition altitude; insert a transition pre-point before entering each plot and set it as the transition altitude, and insert a transition post-point after leaving the plot and set it as the transition altitude, so that the cross-plot movement segment flies at a safe altitude and clearly separates the transition segment from the survey segment.

[0077] Regarding the entry and exit points and the reserved distance on the outside: The entry and exit points do not fall on the boundary of the target surface, but are set at a reserved distance outside the target surface along the flight direction. The purpose of setting this reserved distance is to provide a field of view setback space for oblique photography; when the gimbal is adjusted to the tilt angle, the intersection point of the camera's line of sight will be displaced forward or backward. This reserved distance can compensate for the displacement, so that the outermost boundary facade of the image patch (such as the side wall of a building, the outer fence of an illegal building, etc.) can be within the camera's optimal evidence-gathering depth of field and effective field of view, thereby completing the conversion from two-dimensional orthophoto coverage to three-dimensional facade evidence gathering.

[0078] The retreat margin is determined based on the near-ground offset of the camera's view frustum in an tilted orientation. Specifically, let the current relative flight altitude of the UAV be... H The tilt angle of the gimbal (the angle deviating from the vertical downward viewing direction) is: The effective field of view (FOV) of the camera perpendicular to the sweep direction is: Under the assumption of flat terrain, the point on the ground projected from the near edge of the camera's tilted view frustum is the horizontal offset distance relative to the drone's flight path point directly below it. d It can be derived from geometric trigonometric relationships as follows:

[0079] in, This indicates the horizontal offset distance of the bottom point. Indicates relative flight altitude. Indicates the tilt and pitch angle of the gimbal. This indicates the effective field of view. To ensure the edges of the map patch are stably enclosed and sufficient safety margin is provided, a setback distance is reserved at the final entry / exit points. D Set as:

[0080] in, This indicates that a distance has been reserved. This indicates the minimum safe outward expansion threshold set by the system (to prevent insufficient backing at low altitudes or small tilt angles). This represents the boundary redundancy compensation amount, used to offset the GPS positioning errors of the UAV and the digitization errors of the polygon boundaries of the map features. Preferably, 5.0m is acceptable. 0.5m is acceptable.

[0081] For the drone's tilted forensic posture and photographic sequence: To fully meet the survey requirements for the boundary elevation and overall features of the map patches, and differing from conventional downward-looking orthophotos, the following forensic sequence is configured: 1. Transition Point: Serving as a safe approach point before entering the target patch's airspace, this point primarily handles the transition in spatial position and altitude. The altitude of this point is a preset safe transition altitude. In one embodiment, this point does not trigger a photographing command, and the aircraft yaw angle aligns with the flight path, while the gimbal pitch is initialized or maintained at a normal flight angle (e.g., horizontal forward look) to ensure airspace safety and flight stability during the movement across the patch.

[0082] 2. Entry Point: Calculate the direction vector of this waypoint relative to the interior of the patch (such as the centroid or principal axis), set the drone's heading to point inwards, then adjust the gimbal pitch to the tilt angle and trigger the image capture. This allows for the acquisition of a tilted, panoramic view of the entire patch's edge and interior before the drone enters the survey area.

[0083] 3. Survey Point: Trigger photography in orthogonal mode; when the survey point is immediately following the entry point, the gimbal can be calibrated to orthogonal angle before photography; furthermore, if it is an edge survey point on a regular patch reciprocating scan route, control its fuselage yaw angle to conduct opposing observations along the axis of the current scan line (including forward and reverse directions), and combine this with the gimbal's downward tilting motion to obtain the elevation features of the plot boundary along the scan line direction. 4. Exit point: Determine the incoming heading based on the direction of movement away from the point, set the aircraft heading to face away from the flight direction and point towards the inside of the patch, then adjust the gimbal pitch to the tilt angle and trigger the photo to complete the "look back" facade evidence collection before leaving the survey area.

[0084] 5. Transition point: Raise or maintain the flight altitude to the transition altitude, and return the gimbal pitch to the normal flight angle to reset the status for flying to the next patch.

[0085] When the panoramic viewpoint is enabled, the panoramic viewpoint of a small patch can be placed at the same coordinates as its unique orthophoto point and inserted after it; the panoramic viewpoint of a non-small patch can be placed at the end of the patch and before the transition point, and the transition point and the panoramic viewpoint can be set to the same latitude and longitude so that the transition section can be directly entered after panoramic shooting.

[0086] When the input parameters provide Digital Elevation Model (DEM) data and terrain-following flight mode is enabled, an airspace safety check and interpolation correction are performed on the line segments connecting any adjacent waypoints in the global discrete waypoint sequence: 1. Terrain discrete sampling: On the line segment connecting adjacent waypoints, equidistant sampling is performed according to a preset terrain sampling step size to obtain a set of sampling points; in one embodiment, the terrain sampling step size is denoted as d, which is required to be greater than 0 and can be 10.0m; 2. Route elevation interpolation: For each sampling point, linear interpolation is performed based on the spatial three-dimensional coordinates of the waypoints at both ends of the connecting segment to calculate the absolute elevation interpolation value of the sampling point on the predetermined track; 3. Clearance safety calculation: Read the terrain elevation value corresponding to the sampling point location in the elevation model, and subtract the absolute height interpolation from the terrain elevation value to obtain the actual flight clearance at the sampling point; 4. Adaptive Terrain Interpolation and Lifting: If a sampling point is detected where the actual flight clearance is less than the preset safe clearance height, the spatial location of the minimum clearance value within the connecting line segment is extracted, and a terrain safety transition point is inserted at this point. The relative ground height of the terrain safety transition point is raised to the orthophoto coverage flight height (i.e., its absolute height is recalculated as the sum of the terrain elevation and the orthophoto flight height), thereby using the newly formed broken line segment to bypass the terrain protrusion and meet the clearance safety constraints. In one embodiment, the safe clearance height can be 50.0m, and it is required to be not less than 0, while not exceeding the orthophoto coverage flight height and the transition height; otherwise, an out-of-bounds anomaly will be output.

[0087] Furthermore, the terrain safety transition points serve only as spatial navigation nodes for avoiding terrain obstacles and do not trigger operational actions such as gimbal rotation or image acquisition. Therefore, in the generation of discrete waypoint sets, their arrival action sequences are left empty and configured as pure navigation and safety transition roles to facilitate differentiation at the rendering end and ensure compatibility with the underlying flight control protocol during the output stage. The airspace safety correction mechanism relies on DEM data reflecting terrain undulations and does not avoid vegetation or man-made structures not represented in the elevation model.

[0088] In addition, after generating the basic waypoint sequence, an adaptive coverage repair mechanism based on a minimum cost insertion strategy is also included, specifically including: 1. Footprint assessment and gap extraction: Calculate the union of ground cover footprints of the current waypoint sequence, and extract the set of missing coverage gaps by performing set difference operation with the target surface; 2. Adaptive Local Trajectory Generation: When the area of ​​the missing gap set does not meet the preset tolerance condition, the gap to be repaired is selected based on the preset priority rule; the main axis direction of the reshoot is determined according to the local morphological extension characteristics of the gap to be repaired, and a coherent local reshoot flight point subsequence is adaptively generated based on the main axis direction of the reshoot and the camera coverage parameters. 3. Minimum Cost Global Insertion: Treat the local re-capture waypoint subsequence as an independent insertion unit, and iterate through and calculate the additional flight cost generated by inserting it into each candidate node position in the current basic waypoint sequence; select the optimal node position and corresponding direction that minimizes the additional flight cost and perform the insertion, wherein the additional flight cost is determined based on the change in spatial trajectory length; 4. Iterative closed loop: The footprint assessment and minimum cost global insertion process are executed iteratively until the missing coverage gap meets the preset tolerance condition or reaches the preset repair limit.

[0089] In addition, such as Figure 7 As shown, Figure 7This is a schematic diagram of the ray-based footprint projection in an embodiment of the present invention. In this method, four corner rays of the image plane are constructed based on the pinhole model, and the camera direction basis vector in the world coordinate system is established according to the fuselage heading angle and gimbal pitch angle at the waypoint. The four corner rays are then transformed to the world coordinate system direction. Then, the ray-plane intersection with the ground plane is performed to obtain the four corner points of the ground, forming a ground coverage footprint polygon. The footprint is calculated for waypoints that include the photography action, and the union of the orthophoto footprints is calculated and subtracted from the target surface to obtain the missed coverage area, which is used for debugging visualization and coverage acceptance.

[0090] In practical applications, when the input vector polygon boundary data includes multiple polygon patches, such as... Figure 8 As shown, Figure 8 This is a schematic diagram illustrating the multi-patch task determination and global waypoint concatenation in an embodiment of the present invention. The planning results of each patch are concatenated into a single global discrete waypoint sequence for output, and patch access order information is also output (e.g., a list of patch access order and its identifier sequence is output).

[0091] Determining the order of accessing image patches may include: 1. If the input provides a sequence of identifiers for accessing map features, then the access order of the map features is determined according to the sequence of identifiers. 2. If no map patch access order identifier sequence is provided, the map patch access order is determined using a local optimization strategy based on shortest spatial distance, starting from the takeoff point: In the set of unvisited map patches, the distance between the representative point of each map patch and the current point is calculated, and the map patch with the smallest distance is selected as the next map patch; where the representative point of the map patch is taken as the centroid of the target surface geometry of the map patch; when there is a critical case of equal distance, the map patch with the smaller input index is selected. For each map patch, a survey track and discrete waypoint subsequence are first generated according to the map patch's own algorithm, and the pre-transition point and post-transition point are inserted. When the terrain simulation mode is enabled and DEM data is provided as input, airspace safety checks and insertion point corrections can be performed on the line segments connecting any adjacent waypoints in the global discrete waypoint sequence.

[0092] Finally, to verify the effectiveness of the patch survey, coverage acceptance and omission detection were conducted, such as... Figure 9 As shown, Figure 9 This is a schematic diagram illustrating coverage verification in an embodiment of the present invention, specifically including the following steps: 1. Ray construction: Based on the camera pinhole imaging model, construct the four corner rays of the image plane; 2. Coordinate transformation: Obtain the fuselage heading angle and gimbal pitch angle at the waypoint containing the data acquisition command, and transform the four corner rays to the world coordinate system direction according to the fuselage heading angle and gimbal pitch angle; 3. Footprint generation and comparison: The intersection of the transformed four-corner rays with the ground plane is obtained to obtain the corresponding ground footprint polygons; the union of multiple ground footprint polygons is calculated, and the difference operation is performed with the acceptance area determined by the polygon boundary data to output the missed coverage area as the coverage acceptance result.

[0093] To verify the effectiveness and reliability of this method in real-world, complex business environments, a large-scale batch test was conducted using typical land survey data from a certain region. This test dataset contains 7477 map features to be investigated, covering small map features, narrow road / river corridor map features, and complex agricultural and building map features with highly irregular boundaries.

[0094] This method is used to perform automated route planning and fully discrete waypoint generation on the above 7477 map features, and rigorous coverage verification is performed on all generated routes based on the ray-based footprint calculation model of this method. Figure 10 This is a schematic diagram of the actual map strip route generation result in an embodiment of the present invention, such as... Figure 10 As shown, the test results indicate that: 1. For regular and corridor patches, basic flight paths are generated by convex hull planning surfaces and precise footprint intersection screening is performed in combination with target surfaces, effectively avoiding invalid peripheral flights; 2. For complex map patches with coverage gaps, the adaptive coverage repair mechanism based on minimum cost insertion of the present invention can accurately identify the missed areas and automatically plan local re-shooting waypoints; 3. The final global coverage verification results show that all 7,477 map patches achieved complete coverage of the target acceptance area (0% omission rate), and the effective evidence collection of the building facades was fully guaranteed by controlling the opposing tilt attitude of the entry / exit points and the edges of the flight path.

[0095] This invention also provides a UAV map survey route planning device. The UAV map survey route planning device provided by this invention is described below. The UAV map survey route planning device described below can be referred to in correspondence with the UAV map survey route planning method described above. The device includes: The construction module is used to acquire vector polygon boundary data of the area to be investigated, and to construct target surfaces and planning surfaces with different task semantics and geometric features; The classification module is used to calculate the ground coverage size of a single photo based on camera parameters and ground sampling distance, and to classify the patches. The decision-making module generates categorized routes by adopting a matching route generation strategy based on the determined map patch type. The planning module is used to generate a set of discrete waypoints for the UAV based on the classified flight path, and to perform motion orchestration that decouples the spatial position from the evidence collection attitude.

[0096] In operation, this device first constructs the target surface and the planned surface by projecting the vector polygon boundary data of the area to be investigated onto the working coordinate system. Then, the classification module calculates the single-image coverage size and classifies the patches based on camera parameters and GSD. The decision module then generates classified flight paths based on the patch type and the corresponding flight path generation strategy. Finally, the planning module generates a set of discrete waypoints for the UAV based on the classified flight paths and performs motion choreography that decouples spatial position from evidence-gathering attitude. Through this process, corresponding flight path planning is performed for different types of patches, reducing the UAV's flight and photography time in non-target areas, improving coverage efficiency, and solving the problems of poor patch survey efficiency and accuracy in existing related technologies.

[0097] Figure 11 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 11 As shown, the electronic device may include: a processor 1101, a communication interface 1102, a memory 1103, and a communication bus 1104, wherein the processor 1101, the communication interface 1102, and the memory 1103 communicate with each other via the communication bus 1104. The processor 1101 can call logical instructions in the memory 1103 to execute a UAV map patch survey route planning method, which includes: Obtain vector polygon boundary data of the area to be investigated, and construct target surfaces and planning surfaces with different task semantics and geometric features; The ground coverage size of a single photo is calculated based on camera parameters and ground sampling distance, and the patches are classified. Based on the determined map patch type, a matching route generation strategy is adopted to generate categorized routes; Based on the classification of flight paths, a set of discrete waypoints for the UAV is generated, and motion choreography is performed to decouple the spatial position from the evidence collection attitude.

[0098] Furthermore, the logical instructions in the aforementioned memory 1103 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0099] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is able to execute the UAV map survey route planning method provided by the above methods, the method including: Obtain vector polygon boundary data of the area to be investigated, and construct target surfaces and planning surfaces with different task semantics and geometric features; The ground coverage size of a single photo is calculated based on camera parameters and ground sampling distance, and the patches are classified. Based on the determined map patch type, a matching route generation strategy is adopted to generate categorized routes; Based on the classification of flight paths, a set of discrete waypoints for the UAV is generated, and motion choreography is performed to decouple the spatial position from the evidence collection attitude.

[0100] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the UAV map survey route planning method provided by the above methods, the method comprising: Obtain vector polygon boundary data of the area to be investigated, and construct target surfaces and planning surfaces with different task semantics and geometric features; The ground coverage size of a single photo is calculated based on camera parameters and ground sampling distance, and the patches are classified. Based on the determined map patch type, a matching route generation strategy is adopted to generate categorized routes; Based on the classification of flight paths, a set of discrete waypoints for the UAV is generated, and motion choreography is performed to decouple the spatial position from the evidence collection attitude.

[0101] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0102] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0103] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for planning flight routes in UAV map surveys, characterized in that, include: Obtain vector polygon boundary data of the area to be investigated, and construct target surfaces and planning surfaces with different task semantics and geometric features; The ground coverage size of a single photo is calculated based on camera parameters and ground sampling distance, and the patches are classified. Based on the determined map patch type, a matching route generation strategy is adopted to generate categorized routes; Based on the classified routes, a set of discrete waypoints for the UAV is generated, and action choreography is performed to decouple spatial position from evidence collection attitude.

2. The method for planning flight routes for UAV map surveys according to claim 1, characterized in that, Obtain vector polygon boundary data of the area to be investigated, and construct target surfaces and planning surfaces with different task semantics and geometric features, including: Obtain vector polygon boundary data in the area to be investigated; Transform the vector polygon boundary data from the geographic coordinate system to the working projection coordinate system; In the working projection coordinate system, a planning surface for path generation and a target surface for coverage acceptance are constructed simultaneously.

3. The method for planning flight routes for UAV map surveys according to claim 2, characterized in that, Simultaneously constructing a planning surface for path generation and a target surface for coverage acceptance in the working projection coordinate system includes: Perform geometric repair or normalization on the input polygon, process internal hole features, and perform buffered expansion according to the outward expansion margin; The processing of internal hole features includes removing the internal hole features of polygons and retaining only the outermost envelope boundary, so that the area enclosed by the internal holes is incorporated into the planning surface and the target surface; The target surface is used for calculating the coverage of UAV aerial surveys and for final result acceptance, while the planning surface is used to guide the generation of the underlying UAV flight path.

4. The method for planning flight routes for UAV map surveys according to claim 1, characterized in that, The ground cover size of a single photograph is calculated based on camera parameters and ground sampling distance, and the patches are classified, including: The ground coverage width and height of a single photo are calculated based on the camera parameters and the ground sampling distance, and the flight altitude, strip spacing and heading sampling spacing are determined accordingly. A hierarchical filtering strategy, ranging from low to high computational complexity, is used to classify the patches; the types of patches include small patches, strip or corridor patches, and regular patches.

5. The method for planning flight routes for UAV map surveys according to claim 4, characterized in that, A hierarchical filtering strategy, ranging from low to high computational complexity, is employed for map patch classification, including: If the patch satisfies the criterion of a narrow rectangle based on the minimum circumscribed rotating rectangle, then it is determined whether the patch is a strip or a corridor-type patch. If the patch does not satisfy the narrow rectangle criterion based on the minimum bounding rotation rectangle, but satisfies the initial screening and interception strategy based on the morphological equivalent width, then it is determined whether the patch is a regular patch. If the patch does not meet the initial screening interception strategy based on morphological equivalent width, but meets the fine criteria of corridor skeleton, then the patch is determined to be a strip or corridor-type patch.

6. The method for planning flight routes for UAV map surveys according to claim 4, characterized in that, Based on the determined map patch type, a matching route generation strategy is adopted to generate categorized routes, including: When the patch is determined to be a small patch, it supports generating a basic waypoint sequence from a single orthophoto overlay; When the patch is determined to be a strip or corridor-type patch, the skeleton network is extracted inside the planning surface and a node graph model is constructed. The end-to-end trunk path is obtained through the trunk path search algorithm, and the centerline waypoint sequence is obtained by resampling the end-to-end trunk path. When the patch is determined to be a regular patch, a basic waypoint sequence is generated using a reciprocating scanning method.

7. The method for planning flight routes for UAV map surveys according to claim 6, characterized in that, Extracting the skeleton network within the planning plane and constructing a node graph model includes: The boundary points are sampled at arc length steps to obtain a set of boundary points, and a Voronoi diagram of the boundary points is constructed. The edges of the Voronoi diagram are taken and clipped into the planning plane to obtain a set of candidate skeleton line segments. Non-main line segments with a length less than a preset threshold and small redundant branches that are too close to the boundary are removed to obtain the skeleton line network. The skeleton network is sampled at equal intervals to obtain a set of nodes, and spatially nearest nodes are fused based on a preset distance threshold. Connect nodes with undirected edges according to their adjacent radii, and use Euclidean distance as the edge weight; Each node is limited to retaining a maximum of a preset number of adjacent edges that are closest to the node, and the largest connected component is selected as the main graph of the node graph model.

8. The method for planning flight routes for UAV map surveys according to claim 6, characterized in that, An end-to-end trunk path is obtained through a trunk path search algorithm, and a centerline waypoint sequence is obtained by resampling the end-to-end trunk path, including: In the node graph model, a starting point is selected, and the first shortest path search is performed to obtain the farthest first endpoint. Using the first endpoint as the source point, perform a second shortest path search to obtain the second endpoint that is furthest away; The path between the first endpoint and the second endpoint is reconstructed based on the predecessor relationship of the second shortest path search as the end-to-end backbone path; The cumulative arc length is calculated for the end-to-end trunk path, and an arc length sequence is generated according to a preset sampling interval. Resampled waypoints are obtained by linear interpolation of the arc length on adjacent broken line segments, and a centerline waypoint sequence is generated.

9. The method for planning flight routes for UAV map surveys according to claim 1, characterized in that, Based on the classified flight paths, a set of discrete waypoints for the UAV is generated, and motion choreography that decouples spatial position from evidence-gathering attitude is performed, including: Assign waypoint role identifiers to waypoints in the discrete waypoint set and configure arrival action sequences corresponding to the waypoint role identifiers; the arrival action sequences include at least tilt observation and forward / backward tilt observation attitude control for map boundaries and specific waypoint roles, in order to obtain the three-dimensional elevation features of the target features; The output includes the three-dimensional spatial location and the sequence of actions to the point, and controls the UAV to perform the three-dimensional map evidence collection task of the area to be investigated.

10. A UAV map survey route planning device, characterized in that, include: The construction module is used to acquire vector polygon boundary data of the area to be investigated, and to construct target surfaces and planning surfaces with different task semantics and geometric features; The classification module is used to calculate the ground coverage size of a single photo based on camera parameters and ground sampling distance, and to classify the patches. The decision-making module generates categorized routes by adopting a matching route generation strategy based on the determined map patch type. The planning module is used to generate a set of discrete waypoints for the UAV based on the classified routes, and to perform motion orchestration that decouples spatial position from evidence collection attitude.

Images