Air-ground autonomous cooperative inspection method and device

Abstract

The invention relates to an air-ground autonomous collaborative inspection method and device, which dynamically generates an inspection environment map according to laser radar and visible light image data acquired by air-ground collaborative equipment, and autonomously decides inspection behaviors in real time and generates and executes an inspection path according to geometric and semantic information of the environment, thereby realizing air-ground autonomous collaborative inspection. According to the invention, in combination with the perception and behavior capabilities of the air-ground equipment, through dynamic mapping and autonomous decision making, completely autonomous air-ground cooperative inspection of the inspection area is realized. And meanwhile, key points of the inspection target facility can be extracted and inspected according to the multi-modal sensing data, so that the pertinence and efficiency of the inspection system are improved. The method is applied to the fields of major infrastructure inspection, electric power inspection and the like, and the autonomy and accuracy of unmanned inspection are remarkably improved through autonomous perception, decision making and collaboration.

Description

Technical Field

[0001] The present invention relates to the field of unmanned inspection, and in particular to an air-ground autonomous collaborative inspection method and device. The technology is mainly used in the fields of infrastructure inspection, power inspection, etc. Background Art

[0002] With the rapid development of automation and intelligent technology, unmanned inspection has become a development trend. Improving the efficiency, accuracy and autonomy of air-ground collaborative inspection has become particularly critical. Air-ground collaborative inspection is widely used in major infrastructure inspection, power inspection, farmland monitoring and emergency rescue, especially in rapid inspection in complex and wide-area scenarios. Compared with traditional manual inspection methods, it has the advantages of high efficiency, low cost and high safety.

[0003] However, the existing inspection system relies on a single unmanned platform, which faces challenges in terms of efficiency, stability, and endurance. It is difficult to achieve efficient collaboration and accurate inspections in complex and changing environments. On the one hand, the inspection efficiency and stability of the drone inspection system are limited by its endurance, payload, and wind resistance. On the other hand, the ground robot inspection system has high positioning and navigation accuracy and endurance, but its traffic capacity is limited by the terrain, making it difficult to fully obtain inspection target information.

[0004] In the face of the above challenges, air-ground autonomous and efficient collaborative inspection technology has emerged. By integrating the advantages of drones and unmanned vehicles, and through air-ground collaborative and efficient collaboration and autonomous planning, the autonomy, efficiency and accuracy of the inspection system in wide-area complex inspection scenarios are improved. This technology has broad application prospects in many fields such as major infrastructure inspection, power inspection, farmland monitoring and emergency rescue. Summary of the invention

[0005] In view of the limitations of the existing technology, the present invention proposes an air-ground autonomous and efficient collaborative inspection method and device. This technology achieves more efficient, stable and flexible air-ground autonomous collaborative inspection by fusing data collected by three different types of sensors, namely visible light image data, LiDAR point cloud data, and inertial measurement unit (IMU) measurement information.

[0006] The air-ground autonomous and efficient collaborative inspection method designed by the present invention comprises the following steps: Generate a 3D voxel map using the open space equipment data in the inspection area; Generate a candidate target set, including inspection facility key points identified from the three-dimensional voxel map and sampled viewpoints extracted from the map boundary; Define a gain function, and by evaluating the gain of each target in the candidate target set, assign the most valuable inspection target to each device in the air-ground cluster in real time, and dynamically generate an inspection path. The gain function includes: Geometric gain calculation, including the volume of the frontier cluster and the path length to the optimal viewpoint of the frontier cluster; Semantic gain calculation, adding gain to points with rich semantic information, including key points of structures, insulators of power facilities, etc.; Multi-machine collaborative gain, guiding the robot away from the positions and target points of other robots in the cluster, and tending to go to the target point assigned to itself; Terrain analysis, iteratively maximize the probability of reaching the inspection target, dynamically decide the current action strategy, generate the motion trajectory, and control the robot to stably and efficiently go to the target point. Furthermore, obtain the visible light image data, lidar data and IMU pose information of the air-ground equipment in the inspection area, and use the multi-machine SLAM method to uniformly convert the air-ground equipment laser point cloud to the map coordinate system, incrementally generate the inspection area point cloud map, and then incrementally generate the three-dimensional voxel map.

[0007] Furthermore, when generating a three-dimensional voxel map, a ray casting method is used to integrate the newly added obstacle information into the three-dimensional voxel map according to the obstacle point cloud data obtained by the air-ground equipment, and all newly detected voxels are marked as free or occupied; the Euclidean signed distance field ESDF value between each voxel and the nearest obstacle is recorded in the three-dimensional voxel map. In the occupied state, the ESDF value is set to 0, and in the free state, the distance to the nearest obstacle is re-searched and calculated as the ESDF value.

[0008] Furthermore, the frontier search and clustering are specifically described as follows: First, during each update iteration, the frontier within the current frame perception range is deleted, and the new frontier is detected and clustered using breadth-first search (BFS). This strategy avoids checking the entire map in each iteration. and the need to accurately compare the states of each voxel, both of which consume a lot of computing power. Subsequently, the frontiers with sizes exceeding the threshold are recursively segmented into frontiers of moderate volume through principal component analysis (PCA) to achieve finer object extraction.

[0009] The viewpoint sampling method is specifically described as follows: viewpoint set is to use specific sensors to cover the frontier The candidate position and direction are used as the current specific inspection target. Around the frontier cluster center A set of viewpoints uniformly sampled in cylindrical coordinates. According to the gain function Evaluate and select the best viewpoint , as the coverage frontier Specific posture.

[0010] Furthermore, the key points of the inspection facilities are extracted from the three-dimensional voxel map using a UPKD method.

[0011] Furthermore, the air-ground collaborative inspection path and behavior are planned autonomously in real time. Define a gain function , identify the most valuable inspection targets, assign inspection targets to each device in the air-ground cluster in real time, and dynamically identify the most valuable exploration targets to promote an efficient collaborative autonomous inspection process. Gain function Taking into account the geometric, semantic, and collaborative gain functions, considering the topological accessibility and semantic information of key inspection points, the air-ground collaborative decision-making inspection behavior mode, and evaluating and selecting the optimal viewpoint. Based on the environmental voxel map, taking into account the air-ground equipment traffic capacity, the inspection path is dynamically generated.

[0012] The gain function calculation process is as follows:

[0013] represents the frontier cluster set, Indicates the status of the robot. represents the state set of all robots, is the information gain of the viewpoint, is the path length; in terms of coordination, Capture the influence of other robots, inspection semantics, Evaluate the inspection semantic value of target points, and key points will get higher scores; , , , is the corresponding positive constant weight.

[0014] Furthermore, the geometric gain calculation is specifically as follows:

[0015]

[0016] in, is the information gain of the viewpoint, is the path length, Indicates viewpoint The frontier cluster volume that a particular sensor can cover, calculated using the ray casting method, Indicates the current position of the robot Go to Viewpoint 3D A-star path.

[0017] The multi-machine collaborative gain is specifically defined as:

[0018] in, represents a set of viewpoints, represents the current target point set of all robots, and are the attractive potential field and the repulsive potential field, respectively. Represents a robot Current location, represents its target point, Indicates that the empty space cluster is Robot target points outside represents the corresponding position set, and Represents the positive constant weight of the corresponding item. Through this item, the robot can be guided away from the position and target point of other robots and tend to go to the target point assigned to itself.

[0019] Furthermore, terrain analysis and trajectory generation can help the unmanned swarm navigate to the target location safely and efficiently. , dynamically decide the current action strategy , generate linear velocity and angular velocity instructions to control the robot to move to the target point stably and efficiently. The specific calculation is:

[0020] Monte Carlo sampling is used to estimate the probability density. Specifically, from Generated Samples , and calculate the probability of reaching the target point based on the unobstructed proportion of the Monte Carlo sampling path:

[0021] in Indicates the current calculation status. represents the corresponding intermediate state, Indicates a path.

[0022] According to the Monte Carlo algorithm, the success probability can be obtained by path sampling, that is, Sampling n paths , calculate the probability of successful arrival respectively , whose mathematical expectation is the state The probability of arrival

[0023]

[0024] Occlusion function Characterization Path The occlusion situation, Indicates that the path is not blocked, otherwise

[0025]

[0026] Therefore, the status The probability of arrival It can be calculated as: .

[0027] Based on the same inventive concept, the present invention also discloses an air-ground autonomous collaborative inspection device, including several unmanned vehicles and drones networked through MESH, wherein the unmanned vehicles and drones are equipped with multi-light sensors, including laser, visible light, and infrared bands; the networked unmanned vehicles and drones perform air-ground autonomous collaborative inspection by an air-ground autonomous collaborative inspection method.

[0028] Based on the same inventive concept, the present invention also discloses a computer-readable medium on which a computer program is stored. When the program is executed by a processor, an air-ground autonomous collaborative inspection method is implemented.

[0029] The present invention designs an air-ground autonomous and efficient collaborative inspection device, which mainly includes two unmanned platforms. The first unmanned platform is a drone, which is a four-rotor drone with the ability to fly autonomously, perform edge computing, and perform mission actions. Develop drone payloads that integrate multi-light sensors such as laser and visible light. The second unmanned platform is a mobile hangar. Based on the unmanned vehicle module, it can perform high-precision positioning and mapping, autonomous path planning in indoor scenes without GNSS conditions, and achieve efficient exploration of indoor spaces. At the same time, it is equipped with a drone take-off and landing platform and a charging module to improve the flexibility and endurance of the drone. Through the MESH self-organizing network, a drone and unmanned vehicle combined system is formed to collaboratively perform perception and mapping, accurately collect multi-source data, and the payload output data contains visible light, corresponding position and other information, and synchronizes the data to the inspection platform in real time, laying a solid equipment and data foundation for the inspection of major facilities.

[0030] The advantages of the present invention are: Design an air-ground collaborative autonomous and efficient inspection method to help make inspection tasks more efficient, unmanned, and automated. For example, the conditional load of mountainous terrain, infrastructure and power facility inspections pose a major challenge to the efficiency and safety of traditional manual inspections, and the air-ground cross-domain inspection method composed of drones and unmanned vehicles can greatly improve the inspection efficiency and safety in challenging scenarios.

[0031] The present invention integrates the advantages of perception, traffic and task execution capabilities of drones and unmanned vehicles. Through efficient autonomous perception, mapping and planning algorithms, it not only overcomes the limitations of each single device, but also provides autonomous inspection solutions with high precision and high reliability in a variety of complex environments. The present invention can completely autonomously complete the air-ground collaborative inspection process without an inspection path and without demarcating inspection sub-areas. First, it can autonomously perceive, map, and draw maps of unknown inspection areas to obtain complete environmental information. Secondly, it can autonomously adjust and plan the inspection targets of each device in real time based on the acquired local environmental geometry, semantic information and the status of the air-ground equipment itself, and can pay attention to the key points of the inspection target facilities. Finally, the path optimization module can autonomously plan the path to the current target point and generate corresponding control instructions to ensure that the air-ground equipment completes the inspection task autonomously, efficiently and safely.

[0032] The air-ground autonomous collaborative efficient inspection device designed by the present invention targets complex inspection environments, focuses on major problems such as precise positioning and autonomous flight of unmanned aerial vehicles, and carries out technical research, breaks through the key technical bottlenecks of multi-modal panoramic perception of unmanned clusters, multi-sensor joint posture state estimation, multi-machine air-ground collaborative positioning and mapping, and autonomous exploration of unmanned systems oriented to task characteristics, couples multiple environmental perception means to recognize indoor spaces, realizes unmanned, autonomous, and collaborative positioning and mapping, establishes a multi-level environmental situation semantic knowledge graph, and forms a technical system for intelligent surveying and remote sensing methods for unmanned clusters in complex indoor and outdoor spaces, which helps inspection personnel work efficiently. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 Flowchart of the air-ground autonomous collaborative and efficient inspection method.

[0034] Figure 2 Flowchart of key inspection target extraction and viewpoint sampling.

[0035] Figure 3 Gain function calculation flow chart.

[0036] Figure 4 Schematic diagram of the air-ground autonomous collaborative efficient inspection device.

[0037] Figure 5 Schematic diagram of autonomous collaborative and efficient inspection of substations between ground and air. DETAILED DESCRIPTION

[0038] The technical solution of the present invention is described below in conjunction with the accompanying drawings and embodiments.

[0039] Embodiment 1 Take the power scenario as an example, see Figure 1 and Figure 5 The air-ground autonomous collaborative inspection method disclosed in this embodiment includes the following steps: Step 1: Obtain visible light image data, lidar data and IMU pose information related to the positioning target in the air-ground collaborative equipment cluster; Step 2: Dynamically generate the inspection environment map, incrementally generate the inspection area point cloud map based on the environmental data obtained by the air-ground collaborative equipment, and then incrementally generate the three-dimensional voxel map. First, based on the environmental laser point cloud obtained by the air-ground equipment at different positions and different perspectives, the multi-machine SLAM method is used to uniformly convert it to the map coordinate system to generate a comprehensive inspection environment point cloud map. Subsequently, the three-dimensional voxel map is calculated from the collaboratively generated point cloud map, which divides the environment into a uniform three-dimensional grid, with each grid being a voxel. Using the ray casting method, the newly added obstacle information is integrated into the three-dimensional voxel map based on the obstacle point cloud data obtained by the air-ground equipment, and all newly detected voxels are marked as free or occupied. There are three states of voxels in the voxel map: unknown, free, and occupied, of which the two non-unknown states are called known. In a dynamic environment, when incrementally building a map, the algorithm adds or removes obstacles in real time. Subsequently, the attached Euclidean Signed Distance Functions (ESDF) are updated accordingly, and each voxel in the 3D voxel map records its distance to the nearest obstacle as the ESDF value. The ESDF value update process is as follows: each voxel whose status is updated to occupied or free is added to two queues named insertQueue and deleteQueue respectively. For the elements in insertQueue, ESDF is set to 0. For each voxel in deleteQueue, the distance to the nearest obstacle is re-searched and calculated as value.

[0040] Step 3: Extract key points of inspection facilities and sample boundary viewpoints. Figure 2 As shown in the figure, the key inspection target extraction method aims to achieve two balanced tasks: one is to autonomously obtain a three-dimensional map of the inspection area, and the other is to patrol the key semantic targets. In the process of acquiring the three-dimensional map, the frontier clusters are generated by clustering the boundary voxels to guide the air-to-ground equipment to autonomously explore the unknown area and build the environmental map of the inspection area in an incremental manner. In terms of patrolling the key semantic targets, the UPKD method is used to extract the key points of the inspection facilities based on visible light images and lidar data, and these key points are processed first. The combination of these two aspects of the target is the candidate target set, which includes 1) the viewpoints sampled around the frontier clusters 2) the semantic key points of the inspection facilities.

[0041] Step 3.1, during the acquisition of the three-dimensional map, the frontier clusters are generated by searching and clustering the boundary voxels. These frontier clusters represent the boundary between the free space and the unknown space, where the free space represents the set of voxels marked as known and unoccupied, which can effectively guide the air-ground collaborative equipment to autonomously explore uncovered areas, such as drones and unmanned vehicles. Specifically, when the air-ground equipment performs an inspection task, the system first identifies the boundary of the unknown space and forms a frontier cluster by spatially clustering the boundary voxels. Subsequently, a cylindrical coordinate system is constructed with the cluster center, and viewpoints are sampled within a certain radius as target point candidates. In each iteration, only the voxels within the perception range of the current frame are updated. This strategy significantly reduces the consumption of computing resources by avoiding a comprehensive inspection of the entire map M in each iteration and accurately comparing the status of each voxel one by one. Only the changed area needs to be updated, skipping the redundant processing of the entire map, thereby improving the operating efficiency of the system and reducing the computational overhead.

[0042] Step 3.2, the UPKD method is divided into two processing stages. In the first stage, the UPKD network processes the point cloud to generate candidate inspection points, which consists of two main parts: the data normalization module and the unsupervised key point detection network (UKD-Net). The data normalization module compresses information according to the symmetric structure of the tower, thereby reducing the instability of key point detection. UKD-Net contains a point transformer layer that uses a self-attention mechanism to extract features from the point cloud. In the second stage, a convex optimization strategy is used to filter and obtain inspection points.

[0043] Step 4: The path and behavior of the air-ground collaborative inspection are autonomously planned and adjusted in real time by the algorithm. To achieve air-ground collaborative inspection, the algorithm assigns inspection targets to each device in the air-ground cluster in real time. The air-ground device selects the most valuable inspection target in real time based on its own position, target position, path length, and the importance of target semantic information. At the same time, in order to achieve effective collaboration, the location and target of other devices are considered, so that the device gives priority to inspection targets in other areas to promote an efficient decentralized autonomous inspection process.

[0044] To achieve this goal, a gain function is defined that comprehensively considers geometric features, semantic information, and multi-machine collaborative benefits. The gain function G can identify the most valuable inspection targets. During the planning process, the system not only takes into account geometric accessibility, but also combines the semantic information of inspection key points to collaboratively decide on the inspection mode of air-to-ground equipment. In addition, by evaluating the gains of different viewpoints, the system can select the best observation position and posture. Based on the voxel map of the environment and combined with the traffic capacity of air-to-ground equipment, the system dynamically generates an inspection path. The gain function calculation process is as follows: Figure 3 shown.

[0045]

[0046] represents the frontier cluster set, represents the state set of all robots, Indicates the status of the robot. is the information gain of the viewpoint, is the path length. In terms of collaboration, Capture the impact of other robots. In terms of inspection semantics, The inspection semantic value of the target point is evaluated, and the key point has a higher score. , , , is the corresponding positive constant weight.

[0047] Step 4.1, geometric gain calculation. The geometric gain usually considers the volume of the frontier cluster and the path length to the optimal viewpoint of the frontier cluster. The specific calculation is as follows:

[0048]

[0049] in Indicates viewpoint The frontier cluster volume that a particular sensor can cover, calculated using the ray casting method. Indicates the current position of the robot Go to Viewpoint 3D A-star path. The A-star algorithm is a heuristic algorithm for graph search, widely used in path planning and graph traversal, and can efficiently find the shortest path from the starting point to the target point. 3D A-star refers to the version of the A-star algorithm in three-dimensional space.

[0050] Step 4.2, semantic gain calculation, semantic gain considers the semantic value of the target point, such as key points of the building structure of major infrastructure, insulators of power facilities, etc., and adds gain to points with rich semantic information. Step 4.3, multi-machine collaborative gain, if we consider the robot , given the other members of the team The position and goal of each robot, which is defined as:

[0051] in, represents a set of viewpoints, represents the current target point set of all robots, and are the attractive potential field and the repulsive potential field, respectively. Represents a robot Current location, represents its target point, Indicates that the empty space cluster is Robot target points outside Represents the corresponding position set. Through this item, the robot can be guided away from the positions and target points of other robots, and tend to go to the target point assigned to itself, achieving effective air-ground collaborative autonomous inspection. In addition, and represents the positive constant weight of the corresponding term.

[0052] Step 5: Ensure that the drone cluster can safely and efficiently navigate to the target location through terrain analysis and trajectory generation. The system iteratively optimizes the current action strategy to maximize the probability of reaching the target viewpoint. , thereby generating linear velocity and angular velocity commands to control the robot to move to the target point robustly and efficiently.

[0053] According to the conditional probability formula, the probability of success It can be calculated as the conditional probability product The integral of Indicates the current calculation status. represents the corresponding intermediate state,

[0054] According to the Monte Carlo algorithm, the above probability can be obtained by path sampling, that is, Sampling n paths , calculate the probability of successful arrival respectively , whose mathematical expectation is the state The probability of arrival

[0055]

[0056] Occlusion function Characterization Path The occlusion situation, Indicates that the path is not blocked, otherwise

[0057]

[0058] Therefore, the status The probability of arrival It can be calculated as: .

[0059] Embodiment 2 Based on the same inventive concept, this embodiment discloses an air-ground autonomous collaborative inspection device, as shown in the attached Figure 4 As shown, it includes several unmanned vehicles and drones networked through MESH, and the unmanned vehicles and drones are equipped with multi-light sensors, including laser, visible light, and infrared bands; the networked unmanned vehicles and drones perform air-ground autonomous collaborative inspections by an air-ground autonomous collaborative inspection method.

[0060] Since the device introduced in the second embodiment of the present invention is a device used to implement the method for autonomous air-ground collaborative inspection in the first embodiment of the present invention, a person skilled in the art can understand the specific structure and deformation of the electronic device based on the method introduced in the first embodiment of the present invention, so it is not described here in detail. All electronic devices used in a method in the first embodiment of the present invention belong to the scope of protection of the present invention.

[0061] Embodiment 3 Based on the same inventive concept, the present invention further provides a computer-readable medium on which a computer program is stored, and when the program is executed by a processor, the method described in the first embodiment is implemented.

[0062] Since the device introduced in the third embodiment of the present invention is a computer-readable medium used to implement the method for autonomous air-ground collaborative inspection in the first embodiment of the present invention, a person skilled in the art can understand the specific structure and deformation of the electronic device based on the method introduced in the first embodiment of the present invention, so it is not described here in detail. All electronic devices used in a method in the first embodiment of the present invention belong to the scope of protection of the present invention.

[0063] The specific implementations described herein are merely examples of the spirit of the present invention. Those skilled in the art may make various modifications or additions to the specific implementations described or replace them in similar ways without departing from the spirit of the present invention or exceeding the scope defined by the appended claims.

Claims

1. An air-ground autonomous collaborative inspection method, characterized in that: The following steps are involved: Generate a 3D voxel map using the open space equipment data in the inspection area; Generate a candidate target set, including inspection facility key points identified from the three-dimensional voxel map and sampled viewpoints extracted from the map boundary; Define a gain function, and by evaluating the gain of each target in the candidate target set, assign the most valuable inspection target to each device in the air-ground cluster in real time, and dynamically generate an inspection path. The gain function includes: Geometric gain calculation, including the volume of the frontier cluster and the path length to the optimal viewpoint of the frontier cluster; Semantic gain calculation, adding gain to points with rich semantic information, including key points of structures and insulators of power facilities; Multi-machine collaborative gain, guiding the robot away from the positions and target points of other robots in the cluster, and tending to go to the target point assigned to itself; Analyze terrain, iteratively maximize the probability of reaching the inspection target, dynamically decide the current action strategy, generate motion trajectory, and control the robot to the target point.

2. The air-ground autonomous collaborative inspection method according to claim 1 is characterized in that: Obtain visible light image data, lidar data and IMU pose information of air-ground equipment in the inspection area, and use the multi-machine SLAM method to uniformly convert the laser point cloud of air-ground equipment to the map coordinate system, incrementally generate a point cloud map of the inspection area, and then incrementally generate a three-dimensional voxel map.

3. The air-ground autonomous collaborative inspection method according to claim 2 is characterized in that: When generating a three-dimensional voxel map, the ray casting method is used to integrate the newly added obstacle information into the three-dimensional voxel map according to the obstacle point cloud data obtained by the air-ground equipment, and all newly detected voxels are marked as free or occupied; the Euclidean signed distance field ESDF value between each voxel and the nearest obstacle is recorded in the three-dimensional voxel map. In the occupied state, the ESDF value is set to 0, and in the free state, the distance to the nearest obstacle is re-searched and calculated as the ESDF value.

4. The air-ground autonomous collaborative inspection method according to claim 1 is characterized in that: The key points of the inspection facilities are extracted from the three-dimensional voxel map using the UPKD method.

5. The air-ground autonomous collaborative inspection method according to claim 1 is characterized by: The sampling viewpoint generates a frontier cluster by searching and clustering the boundary voxels of the three-dimensional voxel map, constructs a cylindrical coordinate system with the cluster center, and obtains the sampling viewpoint within a set radius.

6. The air-ground autonomous collaborative inspection method according to claim 1 is characterized by: The gain function calculation process is as follows: represents the frontier cluster set, Indicates the status of the robot. represents the state set of all robots, Indicates the viewpoint, is the information gain of the viewpoint, is the path length; in terms of coordination, Capture the influence of other robots, inspection semantics, Evaluate the inspection semantic value of target points, and key points will get higher scores; , , , is the corresponding positive constant weight.

7. The air-ground autonomous collaborative inspection method according to claim 6 is characterized by: The geometric gain calculation is specifically as follows: in, is the information gain of the viewpoint, is the path length, Indicates viewpoint The frontier cluster volume that a particular sensor can cover, calculated using the ray casting method, Indicates the current position of the robot Go to Viewpoint 3D A-star path; The multi-machine collaborative gain is specifically defined as: in, represents a set of viewpoints, represents the current target point set of all robots, and are the attractive potential field and the repulsive potential field, respectively. Represents a robot Current location, represents its target point, Indicates that the empty space cluster is Robot target points outside represents the corresponding position set, and Represents the positive constant weight of the corresponding item. Through this item, the robot can be guided away from the position and target point of other robots and tend to go to the target point assigned to itself.

8. The air-ground autonomous collaborative inspection method according to claim 1 is characterized by: The specific process of generating motion trajectories is as follows: Calculated as the product of conditional probabilities The integral of Indicates the current calculation status. represents the corresponding intermediate state, According to the Monte Carlo algorithm, the success probability can be obtained by path sampling, that is, Sampling n paths , calculate the probability of successful arrival respectively , whose mathematical expectation is the state The probability of arrival Occlusion function Characterization Path The occlusion situation, Indicates that the path is not blocked, otherwise Therefore, the status The probability of arrival It can be calculated as: 。 9. An air-ground autonomous collaborative inspection device, characterized in that: It includes several unmanned vehicles and drones networked through MESH, and the unmanned vehicles and drones are equipped with multi-light sensors, including laser, visible light, and infrared bands; the networked unmanned vehicles and drones perform air-ground autonomous collaborative inspection by the air-ground autonomous collaborative inspection method described in any one of claims 1-8.

10. A computer readable medium having a computer program stored thereon, characterized in that: When the program is executed by the processor, the air-ground autonomous collaborative inspection method as described in any one of claims 1-8 is implemented.

Images