Unmanned aerial vehicle real-time path planning method based on dynamic topological graph construction

Abstract

The invention relates to the technical field of unmanned aerial vehicle autonomous navigation, in particular to an unmanned aerial vehicle real-time path planning method based on dynamic topological graph construction, and the method comprises the steps: constructing a lightweight dynamic topological graph through environment change detection and influence domain calculation; and executing global path planning based on the topological graph. According to the method, memory and calculation overhead are remarkably reduced, real-time incremental updating is supported, dynamic obstacles are effectively processed, path planning efficiency and robustness are improved, and the method is suitable for autonomous flight of the unmanned aerial vehicle in a complex environment.

Description

Technical Field

[0001] This invention relates to the field of autonomous navigation technology for unmanned aerial vehicles (UAVs), and in particular to a real-time path planning method for UAVs based on dynamic topology graph construction. Background Technology

[0002] With the widespread application of drones, path planning technology has become crucial for ensuring safe drone flight. Existing global path planning for drones relies on map building technologies mainly divided into two categories: SLAM (Simultaneous Localization and Mapping) methods based on geometric metrics and sparse skeleton map methods based on topological relationships. The former constructs 3D raster or voxel maps using LiDAR or visual sensors, but storage and computational costs increase linearly or even superlinearly with the scale of the environment, making it difficult to meet real-time requirements. The latter abstracts the environment into a node-edge graph structure, but typically relies on offline computation of the topological connected graph, making real-time incremental construction impossible and struggling to handle dynamic obstacles and environmental changes.

[0003] Specifically, the existing technology has the following drawbacks:

[0004] 1. Traditional 3D raster or voxel maps have high storage and computational overhead, which increases linearly or superlinearly with the scale of the environment, resulting in high memory and computing resource requirements.

[0005] 2. Offline calculation of topological connectivity graphs based on global 3D raster or voxel maps cannot achieve real-time incremental construction and is difficult to adapt to dynamic environments.

[0006] 3. The topology graph structure is built incrementally based on the original point cloud data, which cannot effectively handle dynamic obstacles and lacks an environmental change detection mechanism, thus failing to achieve "on-demand updates" and resulting in low path planning efficiency.

[0007] Therefore, there is an urgent need for a map model that can accurately represent large-scale environments, support efficient incremental updates, and have fast path query capabilities to meet the needs of UAVs for continuous and safe flight in complex environments. Summary of the Invention

[0008] The purpose of this invention is to address the shortcomings of existing technologies by providing a real-time path planning method for unmanned aerial vehicles (UAVs) based on dynamic topology graph construction.

[0009] To achieve the above objectives, this invention provides a real-time path planning method for unmanned aerial vehicles (UAVs) based on dynamic topology graph construction, comprising the following stages:

[0010] Phase 1: Dynamic Topology Graph Construction: Real-time environmental data is collected through airborne sensors to construct a local probabilistic grid map. Based on environmental change detection and influence domain calculation, topology graph nodes and edges are generated and updated.

[0011] The second stage involves planning the drone's path from the starting point to the destination using a search algorithm, based on the constructed topology map.

[0012] The dynamic topology graph construction steps include:

[0013] Initialize parameters and construct a local probabilistic raster map;

[0014] Environmental change detection and impact domain calculation: Based on the local probabilistic raster map, identify the rasters whose status has changed, calculate the region index to which the changed rasters belong, and update the region index queue;

[0015] Topology graph node generation: Based on the region index queue, a statistical sampling method is used to determine the spatial state of the region, and geometric subdivision is performed on the occupied region to generate topology graph nodes;

[0016] Topology graph node update: Based on the newly generated node queue and the original node queue, calculate the updated nodes, retained nodes, and inserted nodes;

[0017] Topology graph edge generation: potential edges are generated based on node queues, and valid edges are determined through reachability detection;

[0018] Time-series anchor management: Generate topology graph odometer nodes based on UAV odometer data and insert them into the topology graph structure to enhance connectivity.

[0019] In the step "Initialize parameters and construct local probabilistic raster map", the raster queue... Reference point coordinates Region Index Queue Area size (Larger than the size of the drone), area side length Node queue Historical mileage counting node queue odometer cumulative distance ;

[0020] When constructing a local probabilistic grid map: the UAV collects environmental data in real time through onboard sensors and maintains a local probabilistic grid map centered on the UAV's current location.

[0021] The specific methods for environmental change detection and impact domain calculation are as follows:

[0022] Based on the local probabilistic raster map, the raster information of the changed state is determined, and the global coordinates of the raster are inserted into the queue.

[0023] Traversing the queue The global coordinates of the changing raster are calculated, along with the relative coordinates of the preset reference point and the origin. The calculation formula is as follows:

[0024]

[0025] In the formula, Represents global grid coordinates; Represents the global coordinates of the reference point; Represents relative coordinates;

[0026] Based on predefined region size parameters, the 3D region index to which the changing element belongs is calculated using the following formula:

[0027]

[0028] In the formula, Indicates a region index. Indicates the area size parameter, This represents element-wise division;

[0029] Add the region index to the index queue And perform a deduplication operation.

[0030] The specific method for generating topology graph nodes is as follows:

[0031] Traversing the region index queue ;

[0032] The spatial status determination of a region is achieved using a statistical sampling method, as detailed below:

[0033] According to the region index The upper and lower boundaries of the region are calculated using the following formula:

[0034]

[0035] In the formula, Indicates the lower boundary of the region. Indicates the upper boundary of the region. Represents the Hadamard product;

[0036] Within the region to be determined, a 3D sampling grid is generated at the resolution of the local probability raster map. The formula for calculating the number of grid samples is as follows:

[0037]

[0038]

[0039]

[0040] In the formula, , , This indicates the number of sampling points in each dimension. Indicates the resolution of the local probability map. Indicates the area size;

[0041] The formula for calculating the sampling point location is as follows:

[0042]

[0043]

[0044] In the formula, , , , , , Indicates the coordinates of the sampling point;

[0045] For each sampling point, query its occupancy status in the local probability map, including three states: free, occupied, and unknown.

[0046] If any sampling point within the area is in an occupied state, then the entire area is considered occupied.

[0047] If the area is determined to be occupied, geometric subdivision is recursively performed, as follows:

[0048] With the geometric center of the current region The calculation formula is as follows: (The point is the dividing point.)

[0049] Calculate the current side length of the region Its initial value is

[0050] like Less than or equal to the size of the drone Then, octree subdivision is performed, and the calculation formula for the eight sub-regions is as follows:

[0051]

[0052]

[0053] In the formula, Indicates the first Sub-regions , Indicates the first The binary encoded vectors of each sub-region, where the binary encoded matrix values ​​are:

[0054]

[0055] Perform a regional spatial state determination for each sub-region;

[0056] If the region is determined to be free, then the geometric center of that region will be... As a regular node in the topology graph, it is added to the node queue. .

[0057] The specific method for updating the topology graph nodes is as follows:

[0058] Calculate update node ,in ;

[0059] Calculate the reserved nodes ,in This represents the difference operation;

[0060] Calculate the inserted node .

[0061] The specific method for generating edges in the topological graph is as follows:

[0062] Traversing the node queue Obtain the edge queue to be detected based on its neighbor nodes.

[0063] Get the potential edge queue of the updated node The specific steps are as follows:

[0064] Obtain the region index based on the node. The calculation method is based on predefined region size parameters to calculate the three-dimensional region index to which the changing unit belongs;

[0065] Define the main search direction:

[0066]

[0067] Obtain neighborhood index by traversing the search direction The calculation formula is as follows:

[0068] If the main search results Perform a secondary direction search, defined as follows:

[0069]

[0070] Iterate through the nodes in the neighbor region; if a node is connected to the current node, add it to the edge queue.

[0071] Traverse the edge queue , Perform reachability testing, specifically by doing the following:

[0072] A ray is emitted from any node toward another node;

[0073] Check the grid status of the point on the local probability map step by step along the light ray;

[0074] If any checkpoint is occupied, mark the edge as invalid and delete the neighbor relationships and weights between the nodes.

[0075] If all checkpoints are in a free state, then the edge is marked as valid, with its weight being the distance between nodes, and the neighbor relationship between nodes is also added.

[0076] Traversing the node queue Obtain potential edge queues And perform reachability detection.

[0077] The timing anchor point management method is as follows:

[0078] Obtain the current attitude information of the drone, i.e., odometry data. ;

[0079] calculate With historical odometer node queue distance ;

[0080] like The minimum distance is greater than ,for Generate odometer nodes in the topology map and insert them into the historical odometer node queue. ;

[0081] To insert the odometer node into the topology graph structure, perform the following operations:

[0082] according to Get the region index The calculation method is based on predefined region size parameters to calculate the three-dimensional region index to which the changing unit belongs;

[0083] Get the potential edge queue of the node ;

[0084] Traverse the edge queue Perform reachability detection.

[0085] Among them, the methods for planning the path of the drone from the starting point to the destination using a search algorithm based on the constructed topology map include:

[0086] Conventional search strategy: Use the topology node closest to the drone's starting point and ending point as the mapping node, and use Dijkstra's algorithm to search for the path;

[0087] If the regular search fails, switch to the alternative search strategy: use the topology map odometer node closest to the drone's starting point and ending point as the mapping node, and re-execute the path search.

[0088] This invention presents a real-time path planning method for unmanned aerial vehicles (UAVs) based on dynamic topology graph construction. The method represents the globally occupied grid map as a free-space topology graph structure, significantly reducing memory and computational requirements. The proposed local update mechanism based on changing influence domains significantly reduces the computational cost of maintaining the topology graph structure, meeting the response latency requirements of high-speed flight scenarios. The invention proposes time-anchored odometry nodes to enhance global topology connectivity, ensuring path planning quality even in scenarios with perception interruptions, dynamic occlusion, or narrow passages. The proposed hierarchical topology naturally represents non-convex free spaces such as bifurcations, caves, and narrow passages, significantly compressing the search state scale and improving search efficiency and reachability determination speed. Attached Figure Description

[0089] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0090] Figure 1 This is a schematic diagram of the octree subdivision process of the present invention. Detailed Implementation

[0091] The embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, but should not be construed as limiting the present invention.

[0092] Please see Figure 1 , Figure 1 This is a schematic diagram of the octree subdivision process of the present invention.

[0093] This invention provides a real-time path planning method for unmanned aerial vehicles (UAVs) based on dynamic topology graph construction, comprising the following stages:

[0094] Phase 1: Dynamic Topology Graph Construction: Real-time environmental data is collected through airborne sensors to construct a local probabilistic grid map. Based on environmental change detection and influence domain calculation, topology graph nodes and edges are generated and updated.

[0095] In this specific implementation, the dynamic topology graph construction steps include:

[0096] Initialize parameters and construct a local probabilistic raster map;

[0097] Environmental change detection and impact domain calculation: Based on the local probabilistic raster map, identify the rasters whose status has changed, calculate the region index to which the changed rasters belong, and update the region index queue;

[0098] Topology graph node generation: Based on the region index queue, a statistical sampling method is used to determine the spatial state of the region, and geometric subdivision is performed on the occupied region to generate topology graph nodes;

[0099] Topology graph node update: Based on the newly generated node queue and the original node queue, calculate the updated nodes, retained nodes, and inserted nodes;

[0100] Topology graph edge generation: potential edges are generated based on node queues, and valid edges are determined through reachability detection;

[0101] Time-series anchor management: Generate topology graph odometer nodes based on UAV odometer data and insert them into the topology graph structure to enhance connectivity.

[0102] In the step "Initialize parameters and construct local probabilistic raster map", the raster queue... Reference point coordinates Region Index Queue Area size (Larger than the size of the drone), area side length Node queue Historical mileage counting node queue odometer cumulative distance ;

[0103] When constructing a local probabilistic grid map: The UAV collects environmental data in real time through onboard sensors to maintain a local probabilistic grid map centered on the UAV's current location. The local map is a fixed-size two-dimensional or three-dimensional grid structure centered on the UAV's current location. Each grid cell corresponds to a spatial region in the environment, recording the probability of that region being occupied by obstacles. The state of each grid cell in the map is represented by occupied, free, and unknown, where occupied indicates the presence of obstacles, and unknown indicates that the region has not yet been detected by the sensors. State estimation is performed on each grid cell based on an inverse sensor model: grids along the measurement ray are classified as free regions, the grid at the measurement endpoint is classified as occupied regions, and unobserved regions remain in an unknown state. This map serves as the foundation for the subsequent construction of a dynamic topology graph.

[0104] The specific methods for environmental change detection and impact domain calculation are as follows:

[0105] Based on the local probabilistic raster map, the raster information of the changed state is determined, and the global coordinates of the raster are inserted into the queue.

[0106] Traversing the queue The global coordinates of the changing raster are calculated, along with the relative coordinates of the preset reference point and the origin. The calculation formula is as follows:

[0107]

[0108] In the formula, Represents global grid coordinates; Represents the global coordinates of the reference point; Represents relative coordinates;

[0109] Based on predefined region size parameters, the 3D region index to which the changing element belongs is calculated using the following formula:

[0110]

[0111] In the formula, Indicates a region index. Indicates the area size parameter, This represents element-wise division;

[0112] Add the region index to the index queue And perform a deduplication operation.

[0113] The specific method for generating topology graph nodes is as follows:

[0114] Traversing the region index queue ;

[0115] The spatial status determination of a region is achieved using a statistical sampling method, as detailed below:

[0116] According to the region index The upper and lower boundaries of the region are calculated using the following formula:

[0117]

[0118] In the formula, Indicates the lower boundary of the region. Indicates the upper boundary of the region. Represents the Hadamard product;

[0119] Within the region to be determined, a 3D sampling grid is generated at the resolution of the local probability raster map. The formula for calculating the number of grid samples is as follows:

[0120]

[0121]

[0122]

[0123] In the formula, , , This indicates the number of sampling points in each dimension. Indicates the resolution of the local probability map. Indicates the area size;

[0124] The formula for calculating the sampling point location is as follows:

[0125]

[0126]

[0127] In the formula, , , , , , Indicates the coordinates of the sampling point;

[0128] For each sampling point, query its occupancy status in the local probability map, including three states: free, occupied, and unknown.

[0129] If any sampling point within the area is in an occupied state, then the entire area is considered occupied.

[0130] If the area is determined to be occupied, geometric subdivision is recursively performed, as follows:

[0131] With the geometric center of the current region The calculation formula is as follows: (The point is the dividing point.)

[0132] Calculate the current side length of the region Its initial value is

[0133] like Less than or equal to the size of the drone Then perform octree subdivision, such as Figure 1 As shown, the calculation formulas for the eight sub-regions are as follows:

[0134]

[0135]

[0136] In the formula, Indicates the first Sub-regions , Indicates the first The binary encoded vectors of each sub-region, where the binary encoded matrix values ​​are:

[0137]

[0138] Perform a regional spatial state determination for each sub-region;

[0139] If the region is determined to be free, then the geometric center of that region will be... As a regular node in the topology graph, it is added to the node queue. .

[0140] The specific method for updating the topology graph nodes is as follows:

[0141] Calculate update node ,in ;

[0142] Calculate the reserved nodes ,in This represents the difference operation;

[0143] Calculate the inserted node .

[0144] The specific method for generating edges in the topological graph is as follows:

[0145] Traversing the node queue Obtain the edge queue to be detected based on its neighbor nodes.

[0146] Get the potential edge queue of the updated node The specific steps are as follows:

[0147] Obtain the region index based on the node. The calculation method is based on predefined region size parameters to calculate the three-dimensional region index to which the changing unit belongs;

[0148] Define the main search direction:

[0149]

[0150] Obtain neighborhood index by traversing the search direction The calculation formula is as follows:

[0151] If the main search results Perform a secondary direction search, defined as follows:

[0152]

[0153] Iterate through the nodes in the neighbor region; if a node is connected to the current node, add it to the edge queue.

[0154] Traverse the edge queue , Perform reachability testing, specifically by doing the following:

[0155] A ray is emitted from any node toward another node;

[0156] Check the grid status of the point on the local probability map step by step along the light ray;

[0157] If any checkpoint is occupied, mark the edge as invalid and delete the neighbor relationships and weights between the nodes.

[0158] If all checkpoints are in a free state, then the edge is marked as valid, with its weight being the distance between nodes, and the neighbor relationship between nodes is also added.

[0159] Traversing the node queue Obtain potential edge queues And perform reachability detection.

[0160] The timing anchor point management method is as follows:

[0161] Obtain the current attitude information of the drone, i.e., odometry data. ;

[0162] calculate With historical odometer node queue distance ;

[0163] like The minimum distance is greater than ,for Generate odometer nodes in the topology map and insert them into the historical odometer node queue. ;

[0164] To insert the odometer node into the topology graph structure, perform the following operations:

[0165] according to Get the region index The calculation method is based on predefined region size parameters to calculate the three-dimensional region index to which the changing unit belongs;

[0166] Get the potential edge queue of the node ;

[0167] Traverse the edge queue Perform reachability detection.

[0168] The second stage involves planning the drone's path from the starting point to the destination using a search algorithm, based on the constructed topology map.

[0169] In this specific implementation, based on the constructed topology map, the method for planning the path of the UAV from the starting point to the destination using a search algorithm includes:

[0170] Conventional search strategy: Use the topology node closest to the drone's starting point and ending point as the mapping node, and use Dijkstra's algorithm to search for the path;

[0171] If the regular search fails, switch to the alternative search strategy: use the topology map odometer node closest to the drone's starting point and ending point as the mapping node, and re-execute the path search.

[0172] The standard search strategy is as follows:

[0173] S1. Take the topology node closest to the drone's starting and ending points as the mapping node, and denote the starting node as... The target node is ;

[0174] S2. Initialize parameters from the starting point to the node. The actual shortest distance From node Heuristic distance estimation to the target ,node Total cost ;

[0175] S3. Create a priority queue of nodes to be expanded. The set of visited nodes ;

[0176] S4. Execute Dijkstra's algorithm, specifically the following operations:

[0177] from extract The smallest node is selected as the current processing node;

[0178] If the node is the target node, then extract the path sequence by backtracking its parent node; otherwise, add it to the target node. ;

[0179] Perform neighborhood expansion, which involves traversing all neighboring nodes of the current node and calculating the weights between the nodes. ;

[0180] like Less than neighboring nodes Then update it. And the parent node pointer, and insert it into the priority queue. ;

[0181] If priority queue The target node has been exhausted and still not found. Then, alternative search strategies are executed, including using the topology map odometer node closest to the drone's starting and ending points as the mapping node, and denoting the starting node as... The target node is ;

[0182] Perform the standard search strategy steps S2-S4.

[0183] Example 1: Implementation of Dynamic Topology Graph Construction

[0184] Taking an UAV flying in an indoor environment as an example, an airborne LiDAR collects data in real time to construct a local probabilistic grid map (resolution δ=0.1m). The region size S is set to twice the UAV size (e.g., S=1.0m). The environmental change detection cycle is 100ms, updating only the topological nodes within the affected region. When generating topological graph nodes, octree subdivision is used until the sub-region side length E≤0.5m (UAV size). The temporal anchor management threshold is set to 2.0m to ensure connectivity is maintained even in the event of loss of perception.

[0185] Example 2: Implementation of Global Path Planning

[0186] The drone flies from point A to point B, initially searching for a path in the topology map using a conventional search strategy. If the search fails, it switches to an alternative search strategy, using odometry nodes as backup nodes. The path planning cycle is 50ms, meeting real-time requirements.

[0187] Furthermore, during this process, key planning process data, including environmental change grids, topology map update operations, and final paths, are hashed and stored on the blockchain using blockchain technology to ensure the immutability and traceability of the data. Based on this reliable historical data, data mining algorithms (such as cluster analysis to discover typical environmental patterns, or association rule mining to analyze the relationship between specific topology structures and planning success rates) are used for offline analysis to uncover deep association rules between environmental dynamics, topology structures, and path performance, thereby providing feedback to optimize topology map construction parameters and path search strategies.

[0188] This invention provides a real-time path planning method for UAVs based on dynamic topology graph construction. The method represents the globally occupied grid map as a free-space topology graph structure, significantly reducing memory and computational requirements. The proposed local update mechanism based on changing influence domains significantly reduces the computational cost of maintaining the topology graph structure, meeting the response latency requirements of high-speed flight scenarios. The invention proposes time-anchored odometry nodes to enhance global topology connectivity, ensuring path planning quality even in scenarios with perception interruptions, dynamic occlusion, or narrow passages. The proposed hierarchical topology naturally represents non-convex free spaces such as bifurcations, caves, and narrow passages, significantly compressing the search state scale and improving search efficiency and reachability determination speed.

[0189] The above-disclosed embodiments are merely one or more preferred embodiments of this application and should not be construed as limiting the scope of this application. Those skilled in the art can understand that all or part of the processes for implementing the above embodiments and equivalent changes made in accordance with the claims of this application still fall within the scope of this application.

Claims

1. A real-time path planning method for unmanned aerial vehicles (UAVs) based on dynamic topology graph construction, characterized in that, Includes the following stages: Phase 1: Dynamic Topology Graph Construction: Real-time environmental data is collected through airborne sensors to construct a local probabilistic grid map. Based on environmental change detection and influence domain calculation, topology graph nodes and edges are generated and updated. The second stage involves planning the drone's path from the starting point to the destination using a search algorithm, based on the constructed topology map.

2. The real-time path planning method for UAVs based on dynamic topology graph construction as described in claim 1, characterized in that, The steps for constructing a dynamic topology graph include: Initialize parameters and construct a local probabilistic raster map; Environmental change detection and impact domain calculation: Based on the local probabilistic raster map, identify the rasters whose status has changed, calculate the region index to which the changed rasters belong, and update the region index queue; Topology graph node generation: Based on the region index queue, a statistical sampling method is used to determine the spatial state of the region, and geometric subdivision is performed on the occupied region to generate topology graph nodes; Topology graph node update: Based on the newly generated node queue and the original node queue, calculate the updated nodes, retained nodes, and inserted nodes; Topology graph edge generation: potential edges are generated based on node queues, and valid edges are determined through reachability detection; Time-series anchor management: Generate topology graph odometer nodes based on UAV odometer data and insert them into the topology graph structure to enhance connectivity.

3. The real-time path planning method for UAVs based on dynamic topology graph construction as described in claim 2, characterized in that, In the step "Initialize parameters and build local probabilistic raster map", the raster queue Reference point coordinates Region Index Queue Area size (Larger than the size of the drone), area side length Node queue Historical mileage counting node queue odometer cumulative distance ; When constructing a local probabilistic grid map: the UAV collects environmental data in real time through onboard sensors and maintains a local probabilistic grid map centered on the UAV's current location.

4. The real-time path planning method for UAVs based on dynamic topology graph construction as described in claim 3, characterized in that, The specific methods for environmental change detection and impact domain calculation are as follows: Based on the local probabilistic raster map, the raster information of the changed state is determined, and the global coordinates of the raster are inserted into the queue. Traversing the queue The global coordinates of the changing raster are calculated, along with the relative coordinates of the preset reference point and the origin. The calculation formula is as follows: In the formula, Represents global grid coordinates; Represents the global coordinates of the reference point; Represents relative coordinates; Based on predefined region size parameters, the 3D region index to which the changing element belongs is calculated using the following formula: In the formula, Indicates a region index. Indicates the region size parameter, This represents element-wise division; Add the region index to the index queue And perform a deduplication operation.

5. The real-time path planning method for UAVs based on dynamic topology graph construction as described in claim 4, characterized in that, The specific method for generating topology graph nodes is as follows: Traversing the region index queue ; The spatial status determination of a region is achieved using a statistical sampling method, as detailed below: According to the region index The upper and lower boundaries of the region are calculated using the following formula: In the formula, Indicates the lower boundary of the region. Indicates the upper boundary of the region. Represents the Hadamard product; Within the region to be determined, a 3D sampling grid is generated at the resolution of the local probability raster map. The formula for calculating the number of grid samples is as follows: In the formula, , , This indicates the number of sampling points in each dimension. Indicates the resolution of the local probability map. Indicates the area size; The formula for calculating the sampling point location is as follows: In the formula, , , , , , Indicates the coordinates of the sampling point; For each sampling point, query its occupancy status in the local probability map, including three states: free, occupied, and unknown. If any sampling point within the area is in an occupied state, then the entire area is considered occupied. If the area is determined to be occupied, geometric subdivision is recursively performed, as follows: With the geometric center of the current region The calculation formula is as follows: (The point is the dividing point.) Calculate the current side length of the region Its initial value is like Less than or equal to the size of the drone Then, octree subdivision is performed, and the calculation formula for the eight sub-regions is as follows: In the formula, Indicates the first Sub-regions , Indicates the first The binary encoded vectors of each sub-region, where the binary encoded matrix values ​​are: Perform a regional spatial state determination for each sub-region; If the region is determined to be free, then the geometric center of that region will be... As a regular node in the topology graph, it is added to the node queue. .

6. The real-time path planning method for unmanned aerial vehicles (UAVs) based on dynamic topology graph construction as described in claim 5, characterized in that, The specific method for updating topology graph nodes is as follows: Calculate update node ,in ; Calculate the reserved nodes ,in This represents the difference operation; Calculate the inserted node .

7. The real-time path planning method for unmanned aerial vehicles (UAVs) based on dynamic topology graph construction as described in claim 6, characterized in that, The specific method for generating edges in the topological graph: Traversing the node queue Obtain the edge queue to be detected based on its neighboring nodes. Get the potential edge queue of the updated node The specific steps are as follows: Obtain the region index based on the node. The calculation method is based on predefined region size parameters to calculate the three-dimensional region index to which the changing unit belongs; Define the main search direction: Obtain neighborhood index by traversing the search direction The calculation formula is as follows: If the main search results Perform a secondary direction search, defined as follows: Iterate through the nodes in the neighbor region; if a node is connected to the current node, add it to the edge queue. Traverse the edge queue , Perform reachability testing, specifically by doing the following: A ray is emitted from any node toward another node; Check the grid status of the point on the local probability map step by step along the light ray; If any checkpoint is occupied, mark the edge as invalid and delete the neighbor relationships and weights between the nodes. If all checkpoints are in a free state, then the edge is marked as valid, with its weight being the distance between nodes, and the neighbor relationship between nodes is also added. Traversing the node queue Obtain potential edge queues And perform reachability detection.

8. The real-time path planning method for UAVs based on dynamic topology graph construction as described in claim 7, characterized in that, The timing anchor point management method is as follows: Obtain the current attitude information of the drone, i.e., odometry data. ; calculate With historical odometer node queue distance ; like The minimum distance is greater than ,for Generate odometer nodes in the topology map and insert them into the historical odometer node queue. ; To insert the odometer node into the topology graph structure, perform the following operations: according to Get the region index The calculation method is based on predefined region size parameters to calculate the three-dimensional region index to which the changing unit belongs; Get the potential edge queue of the node ; Traverse the edge queue Perform reachability detection.

9. The real-time path planning method for unmanned aerial vehicles (UAVs) based on dynamic topology graph construction as described in claim 8, characterized in that, Based on the constructed topology map, the methods for planning the path of a drone from the starting point to the destination using search algorithms include: Conventional search strategy: Use the topology node closest to the drone's starting point and ending point as the mapping node, and use Dijkstra's algorithm to search for the path; If the regular search fails, switch to the alternative search strategy: use the topology map odometer node closest to the drone's starting point and ending point as the mapping node, and re-execute the path search.

Images