Unmanned vehicle path planning method and device based on orthoimage of unmanned aerial vehicle

Abstract

The invention discloses an unmanned vehicle path planning method and device based on an unmanned aerial vehicle orthoimage, and the method comprises the steps: obtaining an orthoimage shot by an unmanned aerial vehicle, carrying out the semantic segmentation of the orthoimage, and labeling the object types recognized in a semantic segmentation result; extracting a pixel region corresponding to each semantic tag, generating a corresponding semantic category image, and performing TDF truncation Euclidean distance field transformation on each semantic category image; acquiring 3D point cloud data of a laser radar sensor of the unmanned vehicle, mapping the 3D point cloud data of the unmanned vehicle to a coordinate system of the TDF image of the unmanned aerial vehicle by adopting a particle filtering algorithm, acquiring corresponding unmanned vehicle positioning information from the TDF image of the unmanned aerial vehicle, and acquiring obstacle positioning information according to the unmanned vehicle positioning information; and planning the path of the unmanned vehicle according to the unmanned vehicle positioning information and the obstacle positioning information. According to the invention, the positioning precision of the unmanned vehicle and the accuracy and reliability of the navigation path in a complex environment can be improved.

Description

Technical Field

[0001] The present invention relates to the field of unmanned vehicle technology, and in particular to an unmanned vehicle path planning method and device based on unmanned aerial vehicle (UAV) orthophotos. Background Art

[0002] This section is intended to provide a background or context to the embodiments of the invention that are recited in the claims. No statement herein is admitted to be prior art by virtue of its inclusion in this section.

[0003] Drone-assisted navigation for unmanned vehicles (UAVs) is a cutting-edge research area in the field of intelligent systems. Its core approach is to leverage the drone's high-altitude perspective and dynamic perception capabilities to offset the limitations of UAVs' ground-based perspective, thereby enabling collaborative positioning, environmental modeling, and path optimization. Existing technologies primarily implement this approach: UAVs carrying lidar are launched into the air, using SLAM technology to construct point cloud maps. The UAV then uses RTK positioning technology combined with the UAV's point cloud map for navigation. However, existing technologies cannot meet the required precision for UAV positioning and cannot guarantee the accuracy and reliability of the UAV's navigation path. Summary of the Invention

[0004] An embodiment of the present invention provides a method for unmanned vehicle path planning based on drone orthophotos, which is used to improve the positioning accuracy of the unmanned vehicle and ensure the accuracy and reliability of the unmanned vehicle navigation path. The method includes: Obtain orthophotos taken by drones, perform semantic segmentation on the orthophotos to obtain semantic segmentation results, and label the object categories identified in the semantic segmentation results to obtain semantic labels for each object category; object categories include road areas, buildings, vegetation, houses, vehicles, and pedestrians; Extract the pixel area corresponding to each semantic label, generate the corresponding semantic category image, perform TDF truncated Euclidean distance field transform on each semantic category image, and obtain the corresponding TDF image; Obtain the 3D point cloud data from the unmanned vehicle's lidar sensor, use a particle filter algorithm to map the unmanned vehicle's 3D point cloud data to the coordinate system of the drone's TDF image, obtain the corresponding unmanned vehicle positioning information from the drone's TDF image, and obtain the obstacle positioning information based on the unmanned vehicle positioning information; According to the positioning information of the unmanned vehicle and the obstacle positioning information, the path planning of the unmanned vehicle is carried out using the A* algorithm and Bessel curve.

[0005] The embodiment of the present invention further provides an unmanned vehicle path planning device based on drone orthophotos, which is used to improve the positioning accuracy of the unmanned vehicle and ensure the accuracy and reliability of the unmanned vehicle navigation path. The device includes: The UAV orthophoto semantic segmentation module is used to obtain orthophotos taken by drones, perform semantic segmentation on the orthophotos to obtain semantic segmentation results, and label the object categories identified in the semantic segmentation results to obtain semantic labels for each object category; object categories include road areas, buildings, vegetation, houses, vehicles, and pedestrians; The TDF image transformation module is used to extract the pixel area corresponding to each semantic label, generate the corresponding semantic category image, and perform TDF truncated Euclidean distance field transformation on each semantic category image to obtain the corresponding TDF image; The unmanned vehicle positioning information acquisition module is used to obtain the 3D point cloud data of the unmanned vehicle's lidar sensor, map the 3D point cloud data of the unmanned vehicle to the coordinate system of the drone's TDF image using a particle filter algorithm, obtain the corresponding unmanned vehicle positioning information from the drone's TDF image, and obtain obstacle positioning information based on the unmanned vehicle positioning information; The unmanned vehicle path planning module is used to plan the path of the unmanned vehicle based on the unmanned vehicle positioning information and obstacle positioning information, using the A* algorithm and Bessel curve.

[0006] An embodiment of the present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the above-mentioned unmanned vehicle path planning method based on drone orthophotos is implemented.

[0007] An embodiment of the present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the above-mentioned unmanned vehicle path planning method based on drone orthophotos.

[0008] An embodiment of the present invention also provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the above-mentioned unmanned vehicle path planning method based on drone orthophotos.

[0009] In an embodiment of the present invention, an orthophoto taken by a drone is obtained, and semantic segmentation processing is performed on the orthophoto to obtain a semantic segmentation result. The object categories identified in the semantic segmentation result are labeled to obtain semantic labels for each object category; the object categories include road areas, buildings, vegetation, houses, vehicles and pedestrians; the pixel area corresponding to each semantic label is extracted to generate a corresponding semantic category image, and a TDF truncated Euclidean distance field transform is performed on each semantic category image to obtain a corresponding TDF image; 3D point cloud data of an unmanned vehicle lidar sensor is obtained, and a particle filter algorithm is used to map the 3D point cloud data of the unmanned vehicle to the coordinate system of the drone TDF image, and corresponding unmanned vehicle positioning information is obtained from the drone TDF image, and obstacle positioning information is obtained based on the unmanned vehicle positioning information; based on the unmanned vehicle positioning information and the obstacle positioning information, an A* algorithm and a Bessel curve are used to plan the path of the unmanned vehicle. In the above process, the embodiment of the present invention constructs a matching benchmark from the unmanned vehicle's 3D point cloud data to the 2D image through semantic segmentation processing and TDF images of multiple semantic categories, combines the particle filtering algorithm to realize the positioning of the unmanned vehicle's 3D point cloud data to the TDF image, and based on the unmanned vehicle positioning information in the TDF image, uses the A* algorithm and Bessel curve to generate a smooth obstacle avoidance path, thereby improving the positioning accuracy of the unmanned vehicle in complex environments and the accuracy and reliability of the navigation path. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work. In the drawings: Figure 1 Flowchart of a method for unmanned vehicle path planning based on UAV orthophotos in an embodiment of the present invention; Figure 2 Schematic diagram of semantic segmentation of two types of images in an embodiment of the present invention; Figure 3 Schematic diagram of an unmanned vehicle path planning device based on UAV orthophotos in an embodiment of the present invention; Figure 4 Schematic diagram of a path planning device for an unmanned vehicle based on orthophotos of a drone according to an embodiment of the present invention; Figure 5 Schematic diagram of a computer device in an embodiment of the present invention. DETAILED DESCRIPTION

[0011] To make the purpose, technical solutions and advantages of the embodiments of the present invention more clear, the embodiments of the present invention are further described in detail below with reference to the accompanying drawings. Here, the exemplary embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0012] Figure 1 Flowchart of a method for unmanned vehicle path planning based on drone orthophotos in an embodiment of the present invention, the method comprising: Step 101: Obtain an orthophoto taken by a drone, perform semantic segmentation on the orthophoto to obtain a semantic segmentation result, and label the object categories identified in the semantic segmentation result to obtain semantic labels for each object category; the object categories include road areas, buildings, vegetation, houses, vehicles, and pedestrians; Step 102: extract the pixel area corresponding to each semantic label, generate the corresponding semantic category image, perform TDF truncated Euclidean distance field transform on each semantic category image, and obtain the corresponding TDF image; Step 103: Acquire the 3D point cloud data of the unmanned vehicle's lidar sensor, map the 3D point cloud data of the unmanned vehicle to the coordinate system of the drone's TDF image using a particle filter algorithm, obtain the corresponding unmanned vehicle positioning information from the drone's TDF image, and obtain the obstacle positioning information based on the unmanned vehicle positioning information; Step 104 : Based on the unmanned vehicle positioning information and the obstacle positioning information, the path of the unmanned vehicle is planned using the A* algorithm and the Bessel curve.

[0013] Each step is described in detail below.

[0014] In step 101, an orthophoto image taken by a drone is obtained, and semantic segmentation processing is performed on the orthophoto image to obtain a semantic segmentation result. The object categories identified in the semantic segmentation result are labeled to obtain semantic labels for each object category; the object categories include road areas, buildings, vegetation, houses, vehicles, and pedestrians.

[0015] In a specific embodiment, orthophotos generated by images taken by drones are used to achieve navigation of unmanned vehicles. The specific functions of orthophotos are: 1. They are used for cross-perspective positioning of unmanned vehicles; 2. They are used as global static maps for path planning of unmanned vehicles. In the simulation test of the unmanned vehicle positioning algorithm, positioning is achieved by matching the 3D point cloud generated by the lidar on the unmanned vehicle with the 2D orthophotos of the drone without relying on RTK / GPS. In order to improve the accuracy, the orthophotos and lidar point clouds are semantically segmented, the weights of buildings and roads are increased, the weights of trees are reduced, and the impact of the different perspectives of the drone and the unmanned vehicle is weakened; in the path planning process, the orthophotos of the drone provide a global static map for the unmanned vehicle, and the lidar point cloud on the unmanned vehicle detects obstacles in real time to form a local dynamic map, which is combined with the global static map to perform global and local path planning.

[0016] In a specific embodiment, a deep learning model is used to perform semantic segmentation on the orthophotos taken by the drone, and pixel-level semantic labels are identified, and each pixel has a corresponding semantic label. In order to more accurately define the scope of the drivable road, edge detection is also required. The edge detection method can identify boundary lines in images or point clouds. These boundaries are often the dividing lines between roads and the surrounding environment. By extracting these boundary lines, the outline of the drivable road can be further refined. Drone-side operation: In the disaster area, the operator controls a low-cost drone (such as a consumer-grade multi-rotor drone) to take off, take aerial photos of the disaster area, and obtain orthophotos. The image processing module onboard the drone performs semantic segmentation and edge detection on the orthophotos, identifies the boundaries of drivable roads and obstacles, and then generates a TDF image, and transmits the map to the unmanned vehicle through the wireless communication module.

[0017] In step 102, the pixel area corresponding to each semantic label is extracted to generate a corresponding semantic category image, and a TDF truncated Euclidean distance field transform is performed on each semantic category image to obtain a corresponding TDF image.

[0018] Figure 2 Schematic diagram of semantic segmentation of two types of images in an embodiment of the present invention. In a specific embodiment, two types of images are generated based on the semantic segmentation results: For one type of image, the orthophoto is binarized to generate a drivable area map. Roads are set as drivable areas with a value of 255, and areas such as buildings and vegetation are set to 0, providing a global drivable area map for the path planning module.

[0019] In one embodiment, it further includes: Based on the semantic segmentation results and the object categories of the UAV orthophotos, the orthophotos are binarized to generate a binary map of the drivable area: the road areas are marked as drivable areas, and the non-road areas are marked as non-drivable areas.

[0020] Another type of image generates a TDF image (truncated distance field) for each semantic label. There are as many images as there are semantic labels. Taking road images as an example, for semantically segmented orthophotos, the semantic label of each pixel is checked. If it is marked as road, the value is assigned to 255, and if it is marked as other labels, the value is assigned to 0. The TDF image is used to represent the Euclidean distance between any non-target pixel and the target pixel in each semantic category image. The pixel value of the target pixel is the maximum value, and the pixel value of the non-target pixel decreases as the Euclidean distance from the target pixel increases. This Euclidean distance evaluates the degree of match between the unmanned vehicle's lidar 3D point cloud data and the target pixel in the drone's orthophoto.

[0021] In step 103, the 3D point cloud data of the unmanned vehicle's lidar sensor is obtained, and the particle filter algorithm is used to map the 3D point cloud data of the unmanned vehicle to the coordinate system of the drone's TDF image. The corresponding unmanned vehicle positioning information is obtained from the drone's TDF image, and the obstacle positioning information is obtained based on the unmanned vehicle positioning information.

[0022] In one embodiment, a particle filter algorithm is used to map the 3D point cloud data of the unmanned vehicle to the coordinate system of the UAV TDF image, including: Based on the particle state and scale parameters, the unmanned vehicle's lidar 3D point cloud data is mapped to the UAV TDF image.

[0023] In one embodiment, based on particle states and scale parameters, mapping the unmanned vehicle's LiDAR 3D point cloud data to a UAV's TDF image includes: Map the unmanned vehicle's lidar 3D point cloud data to the drone's TDF image using the following mapping formula:

[0024] in, is the spatial coordinate of the particle state, t is the translation amount, is the rotation angle of the unmanned vehicle relative to the TDF image coordinate system for coordinate transformation, is the spatial coordinate of the unmanned vehicle's lidar 3D point cloud data, is the scale parameter, is the pixel coordinate of the mapped TDF image.

[0025] In a specific embodiment, based on the particle state and scale parameters, the spatial coordinates of the LiDAR 3D point cloud data are mapped to the pixels in the TDF image after rotation and translation. The pixel value corresponding to the pixel represents the degree of match between the unmanned vehicle point cloud data and the target pixel. The matching degree is used to calculate the particle weight. The calculation of particle weight in the particle filter algorithm includes the following steps: 1. Determine the degree of semantic matching: Build a semantic feature matching network, input the semantic segmentation feature map of the drone orthophoto and the semantic features of the unmanned vehicle lidar point cloud; output the matching degree of each semantic category s i ; 2. Dynamic weight matrix calculation: based on the degree of matching s i Calculating adaptive semantic weights: For semantic categories with high matching scores (such as buildings in urban scenes), their weights in the particle filter algorithm are increased; for semantic categories with low matching scores (such as vegetation in forest scenes), their weights are reduced. This dynamic weight allocation makes the particle filter's positioning results for the unmanned vehicle more accurate, enabling it to adapt to different scenarios, such as urban and forest scenes, and reducing positioning drift caused by occlusion.

[0026] The weight calculation formula is as follows:

[0027] in, For the The weight of class semantics, For the The degree of matching of class semantics; 3. Particle weight fusion: Generate a weight matrix based on the matching degree of each type of semantics, introduce the weight matrix into the state estimation process of particle filtering, and optimize the positioning results.

[0028] In step 104, the path planning of the unmanned vehicle is performed using the A* algorithm and Bessel curve according to the unmanned vehicle positioning information and the obstacle positioning information.

[0029] In one embodiment, the path planning of the unmanned vehicle is performed using the A* algorithm and Bessel curve based on the unmanned vehicle positioning information and the obstacle positioning information, including: Within the drivable area, the A* algorithm and Bessel curves are used to plan the path of the unmanned vehicle based on its positioning information and obstacle locations. The A* algorithm reduces the number of search nodes while ensuring the shortest path. The Bessel curve ensures the curvature continuity of the local path planning curve. This ensures that the desired trajectory is as smooth as possible while maintaining the shortest path.

[0030] In a specific embodiment, the unmanned vehicle operates as follows: After the unmanned vehicle enters the disaster area, the operator sets the target point for the material drop-off on the orthophoto imagery transmitted by the drone. The unmanned vehicle's lidar acquires real-time point cloud data of the surrounding environment and uses PointNet to perform semantic segmentation, identifying obstacles (such as collapsed buildings and trees) and drivable areas. In the absence of a GPS signal, the unmanned vehicle uses a particle swarm filter algorithm, combining the TDF map and lidar point cloud data for positioning. Based on the positioning results and real-time obstacle information, a cost map is generated. Global and local path planning is then performed using the A* algorithm and Bessel curves. Control signals are then output to control the vehicle's movement, ensuring accurate delivery of materials to the target point.

[0031] In addition, during the path planning process, the dynamic environment prediction model based on the Transformer architecture uses the semantic features of historical lidar point cloud sequences and orthophotos to predict the movement trajectory of obstacles in the future. The specific steps are as follows: 1. Based on the historical LiDAR point cloud sequence, the Transformer encoder extracts spatiotemporal features; the orthophoto semantic information is integrated, and the Transformer decoder outputs the future obstacle motion trajectory prediction; 2. Utilize a reinforcement learning algorithm to obtain the current state space: vehicle position, speed, and obstacle distribution. Based on obstacle trajectory prediction, the weights of the A* algorithm's heuristic function and the adjustment of the Bessel curve control parameters are generated in the action space. The reward function uses the quantitative weighted values ​​of energy consumption, path smoothness, and obstacle avoidance safety to generate the optimal path planning parameters in real time. 3. Generate avoidance paths in advance based on predicted obstacle trajectories, and optimize the paths online during actual driving by combining real-time lidar point cloud data.

[0032] The present invention also provides an unmanned vehicle path planning device based on drone orthophotos, as described in the following embodiments. Because the principles of this device are similar to those of the unmanned vehicle path planning method based on drone orthophotos, the implementation of this device can be referenced to the implementation of the unmanned vehicle path planning method based on drone orthophotos, and any repetitions will not be repeated.

[0033] Figure 3 Schematic diagram of an unmanned vehicle path planning device based on drone orthophotos in an embodiment of the present invention, the device includes: The UAV orthophoto semantic segmentation module 301 is used to obtain orthophotos taken by the UAV, perform semantic segmentation processing on the orthophotos to obtain semantic segmentation results, and label the object categories identified in the semantic segmentation results to obtain semantic labels for each object category; the object categories include road areas, buildings, vegetation, houses, vehicles, and pedestrians; The TDF image transformation module 302 is used to extract the pixel area corresponding to each semantic label, generate the corresponding semantic category image, and perform TDF truncated Euclidean distance field transformation on each semantic category image to obtain the corresponding TDF image; The unmanned vehicle positioning information acquisition module 303 is used to obtain the 3D point cloud data of the unmanned vehicle's lidar sensor, map the 3D point cloud data of the unmanned vehicle to the coordinate system of the drone's TDF image using a particle filter algorithm, obtain the corresponding unmanned vehicle positioning information from the drone's TDF image, and obtain obstacle positioning information based on the unmanned vehicle positioning information; The unmanned vehicle path planning module 304 is used to plan the path of the unmanned vehicle using the A* algorithm and Bessel curve based on the unmanned vehicle positioning information and obstacle positioning information.

[0034] In one embodiment, the unmanned vehicle positioning information acquisition module 303 is further configured to: Based on the particle state and scale parameters, the unmanned vehicle's lidar 3D point cloud data is mapped to the UAV TDF image.

[0035] In one embodiment, the unmanned vehicle positioning information acquisition module 303 is specifically configured to: Map the unmanned vehicle's lidar 3D point cloud data to the drone's TDF image using the following mapping formula:

[0036] in, is the spatial coordinate of the particle state, t is the translation amount, is the rotation angle of the unmanned vehicle relative to the TDF image coordinate system for coordinate transformation, is the spatial coordinate of the unmanned vehicle's lidar 3D point cloud data, is the scale parameter, is the pixel coordinate of the mapped TDF image.

[0037] Figure 4 FIG. 4 is a schematic diagram of a path planning device for an unmanned vehicle based on orthophotos of a UAV according to an embodiment of the present invention. In one embodiment, the device further includes an orthophoto binarization processing module 401, which is specifically used to: Based on the semantic segmentation results and the object categories of the UAV orthophotos, the orthophotos are binarized to generate a binary map of the drivable area: the road areas are marked as drivable areas, and the non-road areas are marked as non-drivable areas.

[0038] In one embodiment, the unmanned vehicle path planning module 304 is specifically configured to: In the drivable area, the path planning of the unmanned vehicle is carried out using the A* algorithm and Bessel curve based on the positioning information of the unmanned vehicle and the obstacle positioning information.

[0039] An embodiment of the present invention further provides a computer device, Figure 5 This is a schematic diagram of a computer device in an embodiment of the present invention. The computer device 500 includes a memory 510, a processor 520, and a computer program 530 stored in the memory 510 and executable on the processor 520. When the processor 520 executes the computer program 530, the unmanned vehicle path planning method based on drone orthophotos is implemented.

[0040] An embodiment of the present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the above-mentioned unmanned vehicle path planning method based on drone orthophotos.

[0041] An embodiment of the present invention also provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the above-mentioned unmanned vehicle path planning method based on drone orthophotos.

[0042] In an embodiment of the present invention, an orthophoto taken by a drone is obtained, and semantic segmentation processing is performed on the orthophoto to obtain a semantic segmentation result. The object categories identified in the semantic segmentation result are labeled to obtain semantic labels for each object category; the object categories include road areas, buildings, vegetation, houses, vehicles and pedestrians; the pixel area corresponding to each semantic label is extracted to generate a corresponding semantic category image, and a TDF truncated Euclidean distance field transform is performed on each semantic category image to obtain a corresponding TDF image; 3D point cloud data of an unmanned vehicle lidar sensor is obtained, and a particle filter algorithm is used to map the 3D point cloud data of the unmanned vehicle to the coordinate system of the drone TDF image, and corresponding unmanned vehicle positioning information is obtained from the drone TDF image, and obstacle positioning information is obtained based on the unmanned vehicle positioning information; based on the unmanned vehicle positioning information and the obstacle positioning information, an A* algorithm and a Bessel curve are used to plan the path of the unmanned vehicle. In the above process, the embodiment of the present invention constructs a matching benchmark from the unmanned vehicle's 3D point cloud data to the 2D image through semantic segmentation processing and TDF images of multiple semantic categories, combines the particle filtering algorithm to realize the positioning of the unmanned vehicle's 3D point cloud data to the TDF image, and based on the unmanned vehicle positioning information in the TDF image, uses the A* algorithm and Bessel curve to generate a smooth obstacle avoidance path, thereby improving the positioning accuracy of the unmanned vehicle in complex environments and the accuracy and reliability of the navigation path.

[0043] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0044] 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 flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, 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 flowcharts and / or block diagrams. 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.

[0045] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work 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 The function specified in one or more boxes.

[0046] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0047] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for unmanned vehicle path planning based on UAV orthophotos, characterized in that: include: Obtain orthophotos taken by drones, perform semantic segmentation on the orthophotos to obtain semantic segmentation results, and label the object categories identified in the semantic segmentation results to obtain semantic labels for each object category; object categories include road areas, buildings, vegetation, houses, vehicles, and pedestrians; Extract the pixel area corresponding to each semantic label, generate the corresponding semantic category image, perform TDF truncated Euclidean distance field transform on each semantic category image, and obtain the corresponding TDF image; Obtain the 3D point cloud data from the unmanned vehicle's lidar sensor, use a particle filter algorithm to map the unmanned vehicle's 3D point cloud data to the coordinate system of the drone's TDF image, obtain the corresponding unmanned vehicle positioning information from the drone's TDF image, and obtain the obstacle positioning information based on the unmanned vehicle positioning information; According to the positioning information of the unmanned vehicle and the obstacle positioning information, the path planning of the unmanned vehicle is carried out using the A* algorithm and Bessel curve.

2. The method according to claim 1, wherein The particle filter algorithm is used to map the 3D point cloud data of the unmanned vehicle to the coordinate system of the UAV TDF image, including: Based on the particle state and scale parameters, the unmanned vehicle's lidar 3D point cloud data is mapped to the UAV TDF image.

3. The method according to claim 2, wherein Based on particle state and scale parameters, the unmanned vehicle's LiDAR 3D point cloud data is mapped to the UAV's TDF image, including: Map the unmanned vehicle's lidar 3D point cloud data to the drone's TDF image using the following mapping formula: in, is the spatial coordinate of the particle state, t is the translation amount, is the rotation angle of the unmanned vehicle relative to the TDF image coordinate system for coordinate transformation, is the spatial coordinate of the unmanned vehicle's lidar 3D point cloud data, is the scale parameter, is the pixel coordinate of the mapped TDF image.

4. The method according to claim 1, wherein Also includes: Based on the semantic segmentation results and the object categories of the UAV orthophotos, the orthophotos are binarized to generate a binary map of the drivable area: the road areas are marked as drivable areas, and the non-road areas are marked as non-drivable areas.

5. The method according to claim 4, wherein Based on the positioning information of the unmanned vehicle and the obstacle positioning information, the A* algorithm and Bessel curve are used to plan the path of the unmanned vehicle, including: In the drivable area, the path planning of the unmanned vehicle is carried out using the A* algorithm and Bessel curve based on the positioning information of the unmanned vehicle and the obstacle positioning information.

6. A path planning device for an unmanned vehicle based on orthophoto images of a drone, characterized in that: include: The UAV orthophoto semantic segmentation module is used to obtain orthophotos taken by drones, perform semantic segmentation on the orthophotos to obtain semantic segmentation results, and label the object categories identified in the semantic segmentation results to obtain semantic labels for each object category; object categories include road areas, buildings, vegetation, houses, vehicles, and pedestrians; The TDF image transformation module is used to extract the pixel area corresponding to each semantic label, generate the corresponding semantic category image, and perform TDF truncated Euclidean distance field transformation on each semantic category image to obtain the corresponding TDF image; The unmanned vehicle positioning information acquisition module is used to obtain the 3D point cloud data of the unmanned vehicle's lidar sensor, map the 3D point cloud data of the unmanned vehicle to the coordinate system of the drone's TDF image using a particle filter algorithm, obtain the corresponding unmanned vehicle positioning information from the drone's TDF image, and obtain obstacle positioning information based on the unmanned vehicle positioning information; The unmanned vehicle path planning module is used to plan the path of the unmanned vehicle based on the unmanned vehicle positioning information and obstacle positioning information, using the A* algorithm and Bessel curve.

7. The device according to claim 6, characterized in that The unmanned vehicle positioning information acquisition module is also used for: Based on the particle state and scale parameters, the unmanned vehicle's lidar 3D point cloud data is mapped to the UAV TDF image.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method according to any one of claims 1 to 5 is implemented.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Images