Unmanned aerial vehicle path planning system and method based on two-way search of obstacle maps

Abstract

The present invention discloses a UAV path planning system and method based on a two-way search of an obstacle map, preprocesses the three-dimensional laser point cloud data within the scope of the power corridor, and unifies the coordinate system of the three-dimensional laser point cloud data according to the navigation and positioning coordinate system used for the UAV flight; segmentation is performed on the preprocessed three-dimensional laser point cloud of the power corridor; projection dimension reduction is performed on the segmented three-dimensional laser point cloud to generate a power corridor obstacle grid map; based on the power corridor obstacle grid map, a two-way random sampling fast expansion random tree algorithm is used to search the UAV path to obtain the UAV power line patrol planning path; the UAV power line patrol planning path is smoothed and simplified to obtain the optimized UAV flight planning path. By constructing a three-dimensional laser point cloud obstacle grid map, the amount of calculation data is reduced, and safe and efficient flight path planning for UAVs is achieved.

Description

Technical Field

[0001] The present invention relates to the field of safe flight of unmanned aerial vehicles and power grid technology, and in particular to a system and method for unmanned aerial vehicle path planning based on two-way search of obstacle maps. Background Art

[0002] Compared with traditional power line inspection methods, drone power line inspection can easily operate in high mountains and deep forests, rivers and lakes where the operation is more difficult. It can better break through the limitations of human inspection and manned helicopter inspection, which is not only more efficient but also safer. Drone power line inspection monitors and checks power failures in real time for transmission lines, improving the safety, efficiency and intelligence level of transmission line inspection. As the technology matures, drones will become a more effective tool for transmission network line inspection. The advantages of drones in rationally allocating resources and improving the overall efficiency of power line inspection will continue to emerge, and they have broad application prospects in the field of safe power grid operation.

[0003] The first thing that needs to be solved for drone power line inspection is the problem of safe flight. Various drone power line inspection systems have the function of line inspection mission planning. How to quickly plan a safe flight path based on the line direction, the distribution of various facilities, and the geographical environment in the line inspection corridor is the primary problem for safe and efficient drone power line inspection. Drone flight path planning usually uses satellite images and topographic maps of the survey area to set a flight altitude higher than the highest altitude of the flight area for route planning. This method has low flight efficiency and poses safety hazards. On the other hand, since the flight altitude is fixed, as the terrain and surrounding corridor environment fluctuate, the ground resolution of the objects being inspected, such as power lines, poles, power facilities, tree barriers and other objects, varies greatly, affecting the accuracy of the line inspection result analysis. Summary of the invention

[0004] The purpose of the present invention is to provide a drone path planning system and method based on a two-way search of an obstacle map, using the existing three-dimensional laser point cloud data of the power corridor to segment and project the three-dimensional point cloud to obtain a grid map of obstacles in the power corridor, and further adopt an improved rapid expansion random tree search path according to the corridor obstacle grid map, and finally smoothly optimize the path to achieve drone line patrol safety path planning and optimization. The present invention converts the three-dimensional laser point cloud of the power corridor into an obstacle grid map, and realizes drone line patrol safety path planning that takes into account the corridor terrain and spatial environment, which can improve the efficiency and safety of drone power line patrol path planning, realize drone line patrol path optimization, and has broad application prospects in the fields of power grid line maintenance, public safety management, disaster prevention and mitigation, etc.

[0005] To achieve this purpose, the UAV path planning system based on two-way search of obstacle map designed by the present invention includes a laser point cloud segmentation processing module, a laser point cloud projection module, and a two-way search path module; the laser point cloud segmentation processing module is used to segment the three-dimensional laser point cloud of the power corridor; the laser point cloud projection module is used to project and reduce the dimension of the three-dimensional laser point cloud after segmentation processing to generate a power corridor obstacle grid map; the two-way search path module is based on the power corridor obstacle grid map generated by the laser point cloud projection module, and adopts a two-way random sampling fast expansion random tree algorithm to perform UAV path search to obtain the UAV power line patrol planning path.

[0006] A method for UAV path planning based on two-way search of obstacle maps comprises the following steps:

[0007] Step 1, segmenting the three-dimensional laser point cloud of the power corridor;

[0008] Step 2, projecting and reducing the dimension of the segmented three-dimensional laser point cloud to generate a grid map of obstacles in the power corridor;

[0009] Step 3: Based on the power corridor obstacle grid map, a bidirectional random sampling rapid expansion random tree algorithm is used to search for UAV paths to obtain a UAV power line patrol planning path.

[0010] The beneficial effects of the present invention are:

[0011] 1) The present invention proposes a set of safe path planning methods and technical processes for UAV power line inspection based on bidirectional search of laser point cloud obstacle grid maps. By constructing a three-dimensional laser point cloud obstacle grid map, the amount of calculation data is reduced, the flight efficiency of the UAV is improved, and safe and efficient flight path planning for the UAV is achieved.

[0012] 2) The present invention performs path planning based on an obstacle grid map that takes into account the three-dimensional terrain of the power corridor and the undulating environment of the surrounding corridor. It can realize the flight path planning of UAVs with variable altitudes, achieve stable ground resolution of UAV power line inspection objects, and improve the accuracy of power line inspection analysis results.

[0013] 3) The bidirectional rapid expansion random tree path search algorithm proposed in this invention can further improve the efficiency and accuracy of UAV path planning. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a system structure diagram of the present invention;

[0015] Figure 2 is a flow chart of the method of the present invention;

[0016] Among them, 1-laser point cloud preprocessing module, 2-laser point cloud segmentation processing module, 3-laser point cloud projection module, 4-bidirectional search path module, 5-path optimization module. DETAILED DESCRIPTION

[0017] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:

[0018] A UAV path planning system based on two-way search of obstacle maps, such as Figure 1 As shown, it includes a laser point cloud preprocessing module 1, a laser point cloud segmentation processing module 2, a laser point cloud projection module 3, a two-way search path module 4 and a path optimization module 5; the laser point cloud preprocessing module 1 is used to preprocess the three-dimensional laser point cloud data within the power corridor, and unify the coordinate system of the three-dimensional laser point cloud data according to the navigation and positioning coordinate system used for the UAV flight to obtain the preprocessed three-dimensional laser point cloud data; the laser point cloud segmentation processing module 2 is used to segment the preprocessed three-dimensional laser point cloud of the power corridor; the laser point cloud projection module 3 is used to project and reduce the dimension of the three-dimensional laser point cloud after segmentation to generate a grid map of obstacles in the power corridor; the two-way search path module 4 is based on the grid map of obstacles in the power corridor generated by the laser point cloud projection module 3, and adopts a two-way random sampling fast expansion random tree algorithm to search the UAV path to obtain the UAV power line patrol planning path; the path optimization module 5 is used to smooth and simplify the UAV power line patrol planning path obtained by the two-way search path module 4 to obtain an optimized UAV flight planning path.

[0019] In the above technical solution, the implementation method of the laser point cloud preprocessing module 1 includes the following steps:

[0020] Step 1.1, collecting the most recent patrol 3D laser point cloud data of the power corridor in the UAV flight area; collecting characteristic indicators of the 3D laser point cloud data including: first, the timeliness of the 3D laser point cloud data does not exceed the set period, which is preferably two years in this embodiment; second, the laser point cloud density of the 3D laser point cloud data is not less than the set point cloud density, which is preferably 1 point / square meter in this embodiment;

[0021] Step 1.2, filtering and denoising the three-dimensional laser point cloud data obtained in step 1.1, removing flying points, and cropping and outputting the three-dimensional laser point cloud data according to the patrol area of ​​the UAV;

[0022] Step 1.3, converting the coordinate system of the three-dimensional laser point cloud data into the UAV flight navigation and positioning coordinate system, which is usually a satellite navigation and positioning coordinate system. The present invention adopts the WGS84 coordinate system.

[0023] In the above technical solution, the laser point cloud segmentation processing module 2 segments and classifies the pre-processed three-dimensional laser point cloud of the power corridor, and marks and identifies the obstacle point cloud objects within the power corridor, laying a foundation for path planning and obstacle avoidance. The specific implementation method includes the following steps:

[0024] Step 2.1, using a voxel grid filtering method to downsample the three-dimensional laser point cloud data output by the laser point cloud preprocessing module 1 to reduce the amount of data and improve data processing efficiency;

[0025] Step 2.2, using the moving least squares method to smooth outlier points isolated from the point cloud body in the three-dimensional laser point cloud data output in step 2.1, thereby removing noise points;

[0026] Step 2.3, performing RANSAC segmentation on the three-dimensional laser point cloud data outputted in step 2.2 to obtain different types of segmented point cloud objects;

[0027] Step 2.4, clustering the segmented point cloud objects output in step 2.3 according to the Euclidean distance to construct a point cloud set of power corridor obstacle objects.

[0028] In the above technical solution, the characteristic index of downsampling the three-dimensional laser point cloud data in step 2.1 of the laser point cloud segmentation processing module 2 is: when the voxel grid filtering method is adopted, the voxel grid centroid is used as the simplification of the voxel laser point cloud; the voxel grid size is the maximum side length of the drone's external bounding box;

[0029] The characteristic indicators used in the RANSAC segmentation process in step 2.3 of the laser point cloud segmentation processing module 2 include: the error tolerance value of the RANSAC algorithm to determine whether the sample meets the model is not less than the set threshold, which is preferably 0.9 in this embodiment;

[0030] The characteristic indicators used in the clustering process in step 2.4 of the laser point cloud segmentation processing module 2 include: first, the distance threshold set for clustering is half of the voxel grid size in the voxel grid filtering method; second, the KD-Tree nearest neighbor search method is used in the clustering process.

[0031] In the above technical solution, the implementation method of the laser point cloud projection module 3 includes the following steps:

[0032] Step 3.1, setting point cloud projection parameters, wherein the point cloud projection parameters include a three-dimensional point cloud range, a grid size, and a projection direction;

[0033] Step 3.2, generating a grid map according to the point cloud projection parameters, wherein all pixels of the grid map are initialized to zero;

[0034] Step 3.3, starting from the pixel in the first row and first column of the grid map, traverse the power corridor point cloud according to the grid range corresponding to the pixel. If there is an obstacle point cloud object in step 2, the pixel is marked as the maximum elevation of the obstacle point cloud;

[0035] Step 3.4, repeat step 3.3 until the grid map pixels are traversed and the grid map of power corridor obstacles is obtained.

[0036] In the above technical solution, the grid size is set to half of the voxel grid size in the voxel grid filtering method; and the projection direction is set to the elevation direction.

[0037] In the above technical solution, the implementation method of the bidirectional search path module 4 includes the following steps:

[0038] Step 4.1, according to the requirements of the drone patrol line, set the starting point and end point of the planned path;

[0039] Step 4.2, according to the point cloud projection parameters set in step 3.1, convert the starting point and the end point of the planned path into the starting point pixel and the end point pixel under the pixel coordinates of the obstacle grid map;

[0040] Step 4.3, using the starting pixel as the seed point, using the random sampling fast expansion random tree algorithm to search the path, the path principle is to avoid obstacles, and the searched path point is used as the starting path node;

[0041] Step 4.4, taking the end point pixel as the seed point, using the fast expansion random tree algorithm to search the path, the path principle is to avoid obstacles, and the searched path point is used as the end point path node;

[0042] Step 4.5, determine whether the starting path node obtained in step 4.3 and the end path node obtained in step 4.4 overlap. The basis for determining whether the nodes overlap is: if the distance between the starting path node and the end path node is greater than 3 pixels and there are no obstacles between the nodes, the nodes are considered to overlap. If the nodes overlap, the path search is terminated.

[0043] In the above technical solution, the path optimization module 5 smoothes and simplifies the UAV power line patrol planning path obtained by the bidirectional search path module 4, and smoothes and simplifies the path on the principle of ensuring that the line connecting the two turning points of the path is a straight line while avoiding all obstacles, so as to realize the straightening of the curve. The specific implementation method includes the following steps:

[0044] Step 5.1, initialize the optimized path P ′ is an empty path;

[0045] Step 5.2: read the UAV power line patrol planning path obtained by the bidirectional search path module 4 as the original path P, and use the starting node m of the original path P as the optimized path P ′ Starting point src ;

[0046] Step 5.3, read the next node m+1 in the original path P, and compare the edge e formed by it and the previous node m m,m+1 Perform a collision check according to the obstacle grid map. If the collision check passes, continue to set m=m+1 to read the node backward; if the collision check fails, add node m+1 to the optimized path P. ′ ;

[0047] Step 5.4, loop through steps 5.2 to 5.3 until the end point is reached and the optimized path P is obtained. ′ .

[0048] In the above technical solution, the collision check in step 5.3 of the specific implementation method of the path optimization module 5 refers to that if the edge e m,m+1 If the distance between the edge and the obstacle in the obstacle grid map is greater than the set threshold, it means that there is no collision and the collision check passes. m,m+1 If the distance between the obstacle and the obstacle in the obstacle grid map is less than the set threshold, it means that there is a collision and the collision check fails;

[0049] The threshold is set according to the size of the drone model, that is, the maximum size of the drone is converted into twice the number of pixels of the grid map.

[0050] A UAV path planning method based on two-way search of obstacle maps, such as Figure 2 As shown, the following steps are included:

[0051] Step 1: pre-process the 3D laser point cloud data within the power corridor, and unify the coordinate system of the 3D laser point cloud data according to the navigation and positioning coordinate system used for UAV flight to obtain the pre-processed 3D laser point cloud data;

[0052] Step 2, segmenting the preprocessed three-dimensional laser point cloud of the power corridor;

[0053] Step 3, projecting and reducing the dimension of the segmented three-dimensional laser point cloud to generate a grid map of obstacles in the power corridor;

[0054] Step 4: Based on the power corridor obstacle grid map, a bidirectional random sampling fast expansion random tree algorithm is used to search for UAV paths to obtain a UAV power line patrol planning path;

[0055] Step 5: Smooth and simplify the UAV power line patrol planning path to obtain an optimized UAV flight planning path.

[0056] The contents not described in detail in this specification belong to the prior art known to those skilled in the art. It should be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0057] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0058] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0059] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0060] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit its protection scope. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that after reading the present invention, those skilled in the art can still make various changes, modifications or equivalent substitutions to the specific implementation methods of the invention, but these changes, modifications or equivalent substitutions are all within the protection scope of the pending claims of the invention.

Claims

1. A UAV path planning system based on two-way search of obstacle maps, characterized by: It comprises a laser point cloud segmentation processing module (2), a laser point cloud projection module (3) and a bidirectional path search module (4); the laser point cloud segmentation processing module (2) is used to segment the three-dimensional laser point cloud of the power corridor; the laser point cloud projection module (3) is used to project and reduce the dimension of the three-dimensional laser point cloud after segmentation processing to generate a grid map of obstacles in the power corridor; the bidirectional path search module (4) uses the grid map of obstacles in the power corridor generated by the laser point cloud projection module (3) as a basis and adopts a bidirectional random sampling fast expansion random tree algorithm to perform a path search for an unmanned aerial vehicle to obtain a planned path for the power line patrol of the unmanned aerial vehicle; the laser point cloud segmentation processing module (2) is used to segment the three-dimensional laser point cloud of the power corridor; the laser point cloud projection module (3) is used to project and reduce the dimension of the three-dimensional laser point cloud after segmentation processing to generate a grid map of obstacles in the power corridor; the bidirectional path search module (4) uses the bidirectional random sampling fast expansion random tree algorithm to perform a path search for an unmanned aerial vehicle to obtain a planned path for the power line patrol of the unmanned aerial vehicle; A preprocessing module (1) and a path optimization module (5); the laser point cloud preprocessing module (1) is used to preprocess the three-dimensional laser point cloud data within the power corridor, and unify the coordinate system of the three-dimensional laser point cloud data according to the navigation and positioning coordinate system used for the UAV flight, so as to obtain the preprocessed three-dimensional laser point cloud data; the three-dimensional laser point cloud data obtained by the laser point cloud preprocessing module (1) is input into the laser point cloud segmentation processing module (2); the path optimization module (5) is used to smooth and simplify the UAV power line patrol planning path obtained by the bidirectional search path module (4), so as to obtain the optimized UAV flight planning path; The implementation method of the laser point cloud segmentation processing module (2) comprises the following steps: Step 2.1, downsampling the three-dimensional laser point cloud data output by the laser point cloud preprocessing module (1) using a voxel grid filtering method; Step 2.2, using the moving least squares method to smooth outlier points isolated from the point cloud body in the three-dimensional laser point cloud data output in step 2.1; Step 2.3, performing RANSAC segmentation on the three-dimensional laser point cloud data outputted in step 2.2 to obtain different types of segmented point cloud objects; Step 2.4, clustering the segmented point cloud objects outputted in step 2.3 according to the Euclidean distance to construct a point cloud set of power corridor obstacle objects; The path optimization module (5) smoothes and simplifies the planned path of the UAV power line inspection obtained by the bidirectional path search module (4), and smoothes and simplifies the path on the principle of ensuring that the line connecting the two turning points of the path is a straight line while avoiding all obstacles; The implementation method of the path optimization module (5) comprises the following steps: Step 5.1, initialize the optimized path P ′ is an empty path; Step 5.2, read the UAV power line patrol planning path obtained by the bidirectional search path module (4) as the original path P, and use the starting node m of the original path P as the optimized path P ′ Starting point src ; Step 5.3, read the next node m+1 in the original path P, and compare the edge e formed by it and the previous node m m,m+1 Perform a collision check according to the obstacle grid map. If it passes, continue to set m=m+1 to read the node backward; if it fails, add node m+1 to the optimized path P. ′ ; Step 5.4, loop through steps 5.2 to 5.3 until the end point is reached and the optimized path P is obtained. ′ ; The implementation method of the laser point cloud preprocessing module (1) comprises the following steps: Step 1.1, collecting the most recent patrol 3D laser point cloud data of the power corridor in the UAV flight area; collecting characteristic indicators of the 3D laser point cloud data including: first, the timeliness of the 3D laser point cloud data does not exceed the set period; second, the laser point cloud density of the 3D laser point cloud data is not less than the set point cloud density; Step 1.2, filtering and denoising the three-dimensional laser point cloud data obtained in step 1.1, removing flying points, and cropping and outputting the three-dimensional laser point cloud data according to the patrol area of ​​the UAV; Step 1.3, converting the coordinate system of the three-dimensional laser point cloud data into the UAV flight navigation positioning coordinate system; The characteristic index of downsampling the three-dimensional laser point cloud data in step 2.1 of the laser point cloud segmentation processing module (2) is: when the voxel grid filtering method is adopted, the voxel grid centroid is used as the simplification of the voxel laser point cloud; the voxel grid size is the maximum side length of the drone's external bounding box; The feature indicators used in the RANSAC segmentation process in step 2.3 of the laser point cloud segmentation processing module (2) include: the RANSAC algorithm determines whether the sample meets the model error tolerance value that is not less than a set threshold; The characteristic indicators used in the clustering process in step 2.4 of the laser point cloud segmentation processing module (2) include: first, the distance threshold set for clustering is half of the voxel grid size in the voxel grid filtering method; second, the KD-Tree nearest neighbor search method is used in the clustering process; The implementation method of the laser point cloud projection module (3) comprises the following steps: Step 3.1, setting point cloud projection parameters, wherein the point cloud projection parameters include a three-dimensional point cloud range, a grid size, and a projection direction; Step 3.2, generating a grid map according to the point cloud projection parameters, wherein all pixels of the grid map are initialized to zero; Step 3.3, starting from the pixel in the first row and first column of the grid map, traverse the power corridor point cloud according to the grid range corresponding to the pixel. If there is an obstacle point cloud object, the pixel is marked as the maximum elevation of the obstacle point cloud; Step 3.4, repeat step 3.3 until the grid map pixels are traversed and the grid map of power corridor obstacles is obtained; The grid size is set to be half of the voxel grid size in the voxel grid filtering method; the projection direction is set to be the elevation direction; The implementation method of the bidirectional search path module (4) comprises the following steps: Step 4.1, according to the requirements of the drone patrol line, set the starting point and end point of the planned path; Step 4.2, according to the point cloud projection parameters set in step 3.1, convert the starting point and the end point of the planned path into the starting point pixel and the end point pixel under the pixel coordinates of the obstacle grid map; Step 4.3, using the starting pixel as the seed point, using the random sampling fast expansion random tree algorithm to search the path, the path principle is to avoid obstacles, and the searched path point is used as the starting path node; Step 4.4, taking the end point pixel as the seed point, using the fast expansion random tree algorithm to search the path, the path principle is to avoid obstacles, and the searched path point is used as the end point path node; Step 4.5, determine whether the starting path node obtained in step 4.3 and the end path node obtained in step 4.4 overlap. The basis for determining whether the nodes overlap is: if the distance between the starting path node and the end path node is greater than 3 pixels and there are no obstacles between the nodes, the nodes are considered to overlap. If the nodes overlap, the path search is terminated.

2. The unmanned aerial vehicle path planning system based on obstacle map bidirectional search according to claim 1, characterized in that: The collision check in step 5.3 of the specific implementation method of the path optimization module (5) refers to that if the edge e m,m+1 If the distance between the edge and the obstacle in the obstacle grid map is greater than the set threshold, it means that there is no collision and the collision check passes. m,m+1 If the distance between the obstacle and the obstacle in the obstacle grid map is less than the set threshold, it means that there is a collision and the collision check fails; The threshold is set to twice the number of pixels of the grid map converted from the maximum size of the drone.

3. A method for UAV path planning based on the system of claim 1, characterized in that: It includes the following steps: Step 1, segmenting the three-dimensional laser point cloud of the power corridor; Step 2, projecting and reducing the dimension of the segmented three-dimensional laser point cloud to generate a grid map of obstacles in the power corridor; Step 3: Based on the power corridor obstacle grid map, a bidirectional random sampling rapid expansion random tree algorithm is used to search for UAV paths to obtain a UAV power line patrol planning path.

4. The method for UAV path planning based on obstacle map bidirectional search according to claim 3 is characterized in that: It also includes the following steps: The input data of step 1 is preprocessed three-dimensional laser point cloud data, and the method for obtaining the preprocessed three-dimensional laser point cloud data is as follows: preprocessing the three-dimensional laser point cloud data within the power corridor, and unifying the coordinate system of the three-dimensional laser point cloud data according to the navigation and positioning coordinate system used for the UAV flight, so as to obtain the preprocessed three-dimensional laser point cloud data; The UAV power line patrol planning path obtained in step 3 is smoothed and simplified to obtain an optimized UAV flight planning path.

5. A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when run, executes the drone path planning method based on two-way search of obstacle maps as described in claim 3 or 4.

Images