Flight safety constraint-based autonomous exploration type patrol path planning method and system for electric unmanned aerial vehicle

Abstract

The application provides a power unmanned aerial vehicle autonomous exploration type inspection path planning method and system based on flight safety constraints, and relates to the unmanned aerial vehicle power inspection field. The path planning method comprises the following steps: acquiring environment information around a power grid equipment area in real time, and constructing a three-dimensional space map; identifying all nodes in the three-dimensional space map, and calculating the probability of each node being an obstacle; identifying the passable area, the prohibited area and the unknown area in the current three-dimensional space map; performing collision detection on each node in the three-dimensional space map, and constructing a rapid exploration random tree; planning an initial exploration path of the unmanned aerial vehicle based on the rapid exploration random tree; and performing path optimization operation on the initial path to obtain a final path. The application adopts the method based on flight safety constraints, plans a smooth path in combination with the rapid exploration random tree, takes into account global and local planning, and brings higher performance and reliability to the autonomous exploration of the unmanned aerial vehicle power inspection.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) path planning, and more particularly to a method and system for autonomous exploration path planning of UAV power line inspection based on flight safety constraints. Background Technology

[0002] To ensure the reliability of power grid operation and prevent large-scale power outages caused by equipment failures, power grid companies conduct regular inspections of transmission lines, substations, and distribution lines. This helps to promptly identify problems and hazards for maintenance, ensuring stable grid operation. Power grid inspection drones, due to their low cost, light weight, small size, and high mobility, have been widely used in the field of power grid inspection. Using drone technology allows for precise fault detection of power grid equipment, reducing manual labor and saving costs while improving the stability and timeliness of power grid inspections, thus ensuring the safe and stable operation of the power grid. In existing technologies, autonomous drone inspections rely on manually creating 3D models beforehand and then offline planning of inspection paths. When the surrounding environment changes or unknown scenarios occur, this technology poses inspection safety risks or may even be unsuitable. Currently, autonomous exploratory path planning for drones is an important research direction in the field of drone inspection. The autonomous exploration task of drones is a key manifestation of their autonomy and intelligence, and path planning, as a fundamental technology for realizing autonomous exploratory drone inspections, has received widespread attention.

[0003] Traditional UAV path planning algorithms mainly include Dijkstra's algorithm based on greedy strategies, A* algorithm based on geometric search, artificial potential field method based on potential field, and Voronoi diagram algorithm based on geometric figures. These algorithms require accurate modeling and description of obstacles in a structured environment and rely on explicit representation of the environment. Another type of path planning algorithm is based on sampling methods, such as PRM and RRT algorithms, which have strong search capabilities and can reduce environmental constraints, but the algorithms have high randomness and computational complexity. Intelligent algorithms such as ant colony optimization and genetic algorithms have also been widely used in path planning, but they have high computational cost and low search efficiency in the later stages.

[0004] Algorithm A, as an excellent representative of traditional path planning algorithms, combines the advantages of other algorithms. It utilizes heuristic information to avoid traversing the entire search space, improving search efficiency while also requiring less computation than intelligent algorithms. Due to its strong portability and high search efficiency, Algorithm A is widely used in grid-based path planning. However, existing Algorithms A still suffer from problems in grid environments, such as numerous polyline paths and oscillations at path transitions. While improved Algorithms A increase search efficiency, they often neglect flight path safety. This can lead to collisions between the UAV and obstacles on the planned path, affecting flight stability and potentially causing serious accidents. Summary of the Invention

[0005] Objective: This invention proposes an autonomous exploratory inspection path planning method for electric unmanned aerial vehicles (UAVs) based on flight safety constraints. Furthermore, it proposes a system for implementing this method. By combining global and local path planning, and comprehensively considering flight safety and search efficiency, the system aims to ensure the UAV's safe exploration in complex environments and efficiently complete autonomous exploration tasks. Introducing collision detection constraints based on aircraft size and obstacle avoidance penalty functions based on node state judgments into the path planning effectively avoids collision risks and makes the planned path safer and more stable, thus effectively solving the aforementioned problems in existing technologies.

[0006] Firstly, a method for autonomous exploratory inspection path planning of electric power drones based on flight safety constraints is proposed, with the following steps:

[0007] S1. Acquire real-time environmental information of the power grid equipment area to construct a three-dimensional spatial map; identify all nodes in the three-dimensional spatial map and calculate the probability that each node is an obstacle;

[0008] S2. Based on the probability of each node being an obstacle calculated in step S1, identify the passable area, the prohibited area, and the unknown area in the current 3D spatial map respectively.

[0009] S3. Perform collision detection on each node in the three-dimensional spatial map and construct a fast exploration random tree;

[0010] S4. Based on the aforementioned fast exploration random tree, plan the initial exploration path for the UAV;

[0011] S5. Perform path optimization operation on the initial path to obtain the final path, and the UAV performs exploration according to the final path.

[0012] In a further embodiment of the first aspect, step S1 involves acquiring environmental information about the power grid equipment area using a three-dimensional lidar sensor and / or a binocular vision sensor.

[0013] In a further embodiment of the first aspect, the process of calculating the probability that each node is an obstacle in step S1 includes:

[0014] S1-1. Construct a three-dimensional spatial map based on an octree. In the three-dimensional spatial map, the time series is recorded as t = 1, 2, 3, ..., T, and the information of each time moment is recorded as i1, i2, i3, ..., iT. Calculate the probability value of the nth node being occupied.

[0015]

[0016] S1-2, Based on the probability value P(n|i) of the nth node being occupied. 1:T Determine the probability that the nth node is an obstacle:

[0017] When the probability value P(n|i) 1:T If ) = 0.5, it indicates that the current node state is unknown; when the probability value P(n|i 1:T If the probability value P(n|i) < 0.5, it indicates that the current node is idle; when the probability value P(n|i) < 0.5, it indicates that the current node is idle. 1:T If )>0.5, it means that the current node is occupied.

[0018] In a further embodiment of the first aspect, step S2 further includes:

[0019] The current node state is represented by cubes of the same size, indicating passable, impassable, and unknown areas in the map environment, as shown in the following expression:

[0020]

[0021] In the formula, m represents a node, and m∈unknow indicates that the current node state is unknown. Based on this node, the unknown region M in the current 3D spatial map is identified. unknow m∈free indicates that the current node is in an idle state, and the passable area M in the current 3D spatial map is identified based on this node. free m∈occupied indicates that the current node is occupied, and the prohibited area M in the current 3D spatial map is identified based on this node. occupied .

[0022] In a further embodiment of the first aspect, step S3 further includes:

[0023] S3-1. Perform collision detection on each node in the three-dimensional spatial map to find the valid sampling point q. sample ;

[0024] S3-2, Determine the valid sampling point q sample Its nearest point q near If the distance D satisfies the threshold condition, output the new node q. new :

[0025] D = ||q sample -q near ||;

[0026]

[0027] In the formula, η min η represents the minimum distance threshold. max This represents the maximum distance threshold.

[0028] In a further embodiment of the first aspect, step S4 further includes:

[0029] S4-1. Construct the evaluation function f(n):

[0030] f(n)=g(n)+α*h(n)+β*Obs(n);

[0031] In the formula, α represents the weight value of the distance from the current node to the target point; Obs(n) represents the obstacle avoidance penalty function of the current node to be selected;

[0032]

[0033]

[0034] In the formula, (x0, y0, z0) represents the coordinates of the starting node, (x i ,y i ,z i (x) represents the coordinates of the current candidate node. g ,y g ,z g () represents the coordinates of the target point;

[0035] The expression for the obstacle avoidance penalty function Obs(n) is as follows:

[0036]

[0037] In the formula, γ is the map resolution; δ is the weight parameter of the idle node; G ocp (n) represents the total number of nodes occupied by candidate node n in the gain calculation model; G free (n) represents the total number of nodes occupied by candidate node n in the gain calculation model;

[0038] S4-2. Using path point n as the center of the gain volume model, traverse the nodes m within the gain model in units of map resolution, determine the node states, and if node m is occupied and observable from path point n, then include it in G. ocp (n); If node m is idle and observable with path node n, then include it in G. free (n):

[0039]

[0040]

[0041] In the formula, m∈unknow indicates that the current node state is unknown, m∈free indicates that the current node state is free, and m∈occupied indicates that the current node state is occupied.

[0042] In a further embodiment of the first aspect, step S5 further includes:

[0043] S5-1, Simplify the drone to a radius of R. UAV The circle whose safe flight distance from obstacles is R. obs Therefore, the radius of the safe zone of the path is:

[0044] R safe =R UAV +R obs ;

[0045] S5-2. Establish a directed line segment L with i as the starting point and i+2 as the ending point. i-＞i+2 A cube V of predetermined size is established with the directed line segment as the center. safe ;

[0046] S5-3, Traverse the cube V based on map resolution. safe The state of the nodes within the path is used to determine whether the path can be directly connected.

[0047]

[0048] In the formula, m∈unknow indicates that the current node state is unknown, m∈free indicates that the current node state is free, and m∈occupied indicates that the current node state is occupied.

[0049] If Status(L) i-＞i+2 If ) = 0, then it represents line segment L. i-＞i+2 If there is an obstacle, retain the path point i+1;

[0050] If Status(L) i-＞i+2 If ) = -1, then it represents line segment L. i-＞i+2 There is an unknown region; retain the path point i+1.

[0051] If Status(L) i-＞i+2 If ) = 1, then it represents line segment L. i-＞i+2 If all the above states are known and there are no obstacles, then discard path point i+1 and directly connect path point i and path point i+2;

[0052] S5-4. Repeat steps S5-1 to S5-3 until the end point of the path is found, and the final path is obtained. The drone will then explore according to the final path.

[0053] In a second aspect, the present invention proposes an autonomous exploratory inspection path planning system for power unmanned aerial vehicles (UAVs), which includes a map building module, a region identification module, a collision detection module, a path planning module, and a path optimization module.

[0054] The map building module is used to acquire real-time environmental information of the power grid equipment area and build a three-dimensional spatial map; it identifies all nodes in the three-dimensional spatial map and calculates the probability that each node is an obstacle;

[0055] Based on the probability that each node is an obstacle, the area identification module identifies the passable areas, prohibited areas, and unknown areas in the current 3D spatial map.

[0056] The collision detection module is used to perform collision detection on each node in the three-dimensional spatial map and construct a fast exploration random tree;

[0057] The path planning module plans the initial exploration path for the UAV based on the fast exploration random tree;

[0058] The path optimization module is used to perform path optimization operations on the initial path to obtain the final path, and the UAV performs exploration according to the final path.

[0059] A third aspect of the present invention provides an electronic device comprising: a processor and a memory storing computer program instructions; wherein the processor, when executing the computer program instructions, implements the autonomous exploratory inspection path planning method for power unmanned aerial vehicles as described in the first aspect.

[0060] In a fourth aspect, the present invention provides a computer-readable storage medium storing at least one executable instruction that, when executed on an electronic device, causes the electronic device to perform the autonomous exploratory inspection path planning method for power drones as described in the first aspect.

[0061] Beneficial effects: The path planning algorithm disclosed in this invention adopts a flight safety constraint-based method, combines Rapid Exploration Random Tree (RRT) and an improved A* algorithm, and introduces an obstacle avoidance penalty function, enabling UAVs to efficiently and safely avoid obstacles and plan smooth paths. At the same time, it improves search efficiency, takes into account both global and local planning, and brings higher performance and reliability to the autonomous exploration of UAV power line inspection. Attached Figure Description

[0062] Figure 1 This is a flowchart of the UAV autonomous exploratory inspection path planning method in an embodiment of the present invention.

[0063] Figure 2 It is an Octomap map built based on radar point cloud data.

[0064] Figure 3 This is a schematic diagram of the RRG node to be connected in an embodiment of the present invention.

[0065] Figure 4This is a schematic diagram comparing path optimization in an embodiment of the present invention.

[0066] Figure 5(a) is a statistical chart of path lengths between the improved A* algorithm and the optimized improved A* algorithm.

[0067] Figure 5(b) is a comparison of path planning time between the improved A* algorithm and the optimized improved A* algorithm.

[0068] Figure 6(a) is a comparison of path planning lengths among the three algorithms in this embodiment of the invention.

[0069] Figure 6(b) is a comparison of path planning time among the three algorithms in the embodiment of the present invention.

[0070] Figure 7 This is a schematic diagram of the structure of the UAV autonomous exploratory inspection path planning system in an embodiment of the present invention.

[0071] Figure 8 This is an internal structural diagram of the electronic device in an embodiment of the present invention. Detailed Implementation

[0072] In the following description, numerous specific details are set forth in order to provide a more thorough understanding of the invention. However, it will be apparent to those skilled in the art that the invention can be practiced without one or more of these details. In other instances, certain technical features well-known in the art have not been described in order to avoid obscuring the invention.

[0073] To address the problems existing in current technologies, this embodiment proposes an autonomous exploratory inspection path planning method for power grid unmanned aerial vehicles (UAVs) based on flight safety constraints. By combining global and local path planning, and comprehensively considering flight safety and search efficiency, this method aims to ensure the safe exploration of UAVs in power grid inspection environments and efficiently complete autonomous exploration tasks. Introducing collision detection constraints based on aircraft size and obstacle avoidance penalty functions based on node state judgments into the path planning effectively avoids collision risks and makes the planned paths of the UAVs safer and more stable.

[0074] The technical solution is described in detail below.

[0075] Example 1:

[0076] See Figure 1 As shown in the figure, this embodiment discloses the specific details of the UAV autonomous exploratory inspection path planning method.

[0077] Step S1: Acquire real-time environmental information of the power grid equipment area to construct a three-dimensional spatial map; identify all nodes in the three-dimensional spatial map and calculate the probability that each node is an obstacle. The specific process is as follows:

[0078] This invention employs a 3D LiDAR sensor to acquire real-time point cloud information of the power grid equipment area environment, and then fuses and updates this information with pre-constructed offline map information of the power grid equipment area environment to build an accurate map environment. It should be noted that the sensor acquiring the power grid equipment area environment information can be a 3D LiDAR sensor or a binocular vision sensor; only the information of the power grid equipment area environment needs to be obtained. A 3D spatial map, octomap, is constructed based on an octree. The global map of octomap is as follows: Figure 2 As shown. In Octomap, probability is used to represent whether a node in space is an obstacle. The probability formula for the nth node being occupied is...

[0079]

[0080] The time series is t = 1, 2, 3, ..., T, and the information recorded at each time point is i1, i2, i3, ..., iT.

[0081] Step S2: Based on the probability of each node being an obstacle calculated in Step S1, identify the passable areas, prohibited areas, and unknown areas in the current 3D spatial map. The specific process is as follows:

[0082] The nodes mentioned in step S1 are hereinafter referred to as voxels. The flowchart of the UAV autonomous exploratory inspection path planning method based on flight safety constraints is as follows: Figure 1 As shown, during autonomous exploration in an unknown environment, a fast exploration random tree is constructed for flight path search by performing collision detection on the voxel states in the map. Voxel states (idle, occupied, unknown) are represented by small cubes of the same size, indicating passable, impassable, and unknown voxel regions in the map environment, respectively:

[0083]

[0084] Each voxel has a unique index, the OctreeKey, which represents its position in the OctoMap. This index consists of integer coordinates on three axes, similar to the form (x, y, z). Each index has a corresponding OctreeNode, which stores the voxel's occupancy status. Therefore, to query the status of a specific sampling point, the 3D position needs to be converted to a 3D point class in the OctoMap library, and then its corresponding index number is retrieved through the OctreeNode to obtain the voxel's status.

[0085] Step S3: Perform collision detection on each node in the 3D spatial map and construct a fast exploration random tree. The specific process is as follows:

[0086] Because the sampling points are highly random and dense, it is necessary to limit the distance between newly grown nodes and existing tree nodes, filter out some redundant points, and reduce the amount of computation. For example... Figure 3 As shown, q is obtained through node voxel state detection. sample Whether a node can become a connection point after being identified as a valid sampling point still needs to be determined by its proximity to the nearest point q. near Does the distance satisfy the threshold constraint?

[0087] q sample With q near The distance between nodes is represented as:

[0088] D = ||q sample -q near ||

[0089] New node q new The location needs to be calculated as follows:

[0090]

[0091] In the formula: η min and η max These represent the minimum and maximum distance thresholds, respectively.

[0092] Step S4: Based on the fast exploration random tree, plan the initial exploration path for the UAV. The specific process is as follows:

[0093] To prevent the drone from drifting during displacement, an obstacle avoidance penalty function is added to the heuristic function of the A* algorithm.

[0094] The expression for the evaluation function f(n) is:

[0095] f(n) = g(n) + α*h(n) + β*Obs(n)

[0096] In the formula: α represents the weight value of the distance from the current node to the target point; Obs(n) represents the obstacle avoidance penalty function of the current node to be selected; the calculation of g(n) and h(n) also needs to be transformed from two-dimensional to three-dimensional space, and the specific calculation content is as follows:

[0097]

[0098]

[0099] In the formula: (x0, y0, z0) represent the coordinates of the starting node, (x i ,y i ,z i (x) represents the coordinates of the current candidate node. g ,y g ,z g() represents the coordinates of the target point.

[0100] The expression for the obstacle avoidance penalty function is as follows:

[0101]

[0102] In the formula: γ is the map resolution; δ is the weighting parameter of the idle voxels; G ocp (n) represents the total number of voxels occupied by the candidate node n in the gain calculation model; G free (n) represents the total number of voxels occupied by the candidate node n in the gain calculation model.

[0103] Using path point n as the center of the gain volume model, traverse the voxels m within the gain model in units of map resolution to determine the voxel state. If voxel m is occupied and observable from path point n, then include it in G. ocp (n), if voxel m is in an idle state and observable with the path point n, then include it in G. free (n).

[0104]

[0105]

[0106] Step S5: Perform path optimization on the initial path to obtain the final path. The UAV then explores according to the final path. The specific process is as follows:

[0107] By introducing the idea of ​​breadth-first search to optimize the path, the drone is simplified to a radius of R. UAV The circle whose safe flight distance from obstacles is R. obs Therefore, the radius of the safe zone of the path is:

[0108] R safe =R UAV +R obs

[0109] Establish a directed line segment L with i as the starting point and i+2 as the ending point. i-＞i+2 Establish a line segment with a width of 2R as the center. safe Length is 2R safe +||L i-＞i+2 ||, A cube V, twice the height of the machine body safe Traversing V based on map resolution safe The voxel state within the path is used to determine whether the path can be directly connected.

[0110]

[0111] Determine the connection status between the two path points i and i+2 based on the above formula. If Status(L i-＞i+2 If ) = 0, then it represents line segment L. i-＞i+2 There are obstacles on the path, so a direct connection cannot be made; remove path point i+1. If Status(L) i-＞i+2 If ) = -1, then it represents line segment L. i-＞i+2 There are unknown areas on the path, so the path cannot be simplified directly; if Status(L) i-＞i+2 If ) = 1, then it represents line segment L. i-＞i+2 The above conditions are all known and there are no obstacles, so we can directly connect the two points to simplify the path. Following this method, we can continue until we find the endpoint of the path, with the following result: Figure 4 As shown.

[0112] Compared to routes generated based on line collision detection, paths optimized based on safe size expansion detection are significantly safer. Although the optimized path is slightly longer than that generated based on line collision detection, it avoids collisions with obstacles, ensuring the safe flight of the drone. Moreover, it is smoother than the original path, eliminating unnecessary redundant points, while also shortening the planned path length and improving flight stability.

[0113] To verify the effectiveness of the improved A* algorithm for UAV obstacle avoidance path planning, simulations were conducted in 3D space using both the traditional A* algorithm and the proposed improved A* algorithm, under autonomous exploration conditions. A real-world scenario was simulated in the environment, and a 20*16*3 map was built using the Gazebo simulation platform within ROS (Robot Operating System). The flight control system was a PX4 equipped with a 16-line LiDAR sensor. Twenty experiments were conducted, focusing on both the improved A* algorithm before and after the improvement, as well as the improved path smoothing. Table 1 lists the parameters used in this path planning experiment.

[0114] Table 1 Path Planning Parameters

[0115]

[0116]

[0117] The optimization of the path planning effect was verified. Figure 5 shows the improved A* algorithm and the optimized path planning effect. Figure 5(a) is a comparison of path lengths, and Figure 5(b) is a comparison of path planning times. The comparison shows that the improved A* algorithm plans a safer and smoother path after optimization.

[0118] Twenty simulation experiments were conducted on the improved A* path planning algorithm, the optimized path, and the path of the traditional algorithm. Figures 6(a) and 6(b) show the recorded path planning time and path length data, respectively. The analysis results are shown in Table 2.

[0119] Table 2 Simulation Data Records

[0120] Traditional autonomous exploration algorithm 66.23 49.42 Improved autonomous exploration algorithm 68.26 50.42 Improved algorithm + optimization 58.75 40.55

[0121] Before optimization, the average path length and planning time were 68.26m and 50.42s, respectively. After optimization, the average path length and planning time were 58.75m and 40.55s, respectively. Compared to the unoptimized original path, the optimized path significantly reduced redundant points, improved path smoothness, shortened the path length by 13.93%, and reduced the path planning time by 19.55%.

[0122] In summary, this invention, based on the background of autonomous exploration in UAV power line inspection, addresses the issue of insufficiently safe and smooth paths planned by traditional autonomous exploration path planning algorithms. It proposes a UAV autonomous exploration inspection path planning method based on flight safety constraints. First, a fast exploration random graph (RRG) is constructed based on voxel states through collision detection. Second, an obstacle avoidance penalty term is added to the evaluation function to optimize the A* algorithm. Finally, redundant points are removed based on the aircraft's safety limits for path optimization. Compared to traditional algorithms, the proposed autonomous exploration path planning algorithm improves path safety by 8.9% and effectively reduces path length by 11.2% and path planning time by 17.9%. This significantly improves UAV flight stability and provides crucial assurance for autonomous exploration in UAV power line inspection.

[0123] Figure 7This is a schematic diagram of a UAV autonomous exploratory inspection path planning system 600 proposed in one embodiment. This system can execute all the processes disclosed in Embodiment 1 above. The system includes a map building module 601, a region identification module 602, a collision detection module 603, a path planning module 604, and a path optimization module 605. The map building module 601 acquires real-time environmental information of the power grid equipment area and constructs a three-dimensional spatial map; it identifies all nodes in the three-dimensional spatial map and calculates the probability that each node is an obstacle. The region identification module 602, based on the probability that each node is an obstacle, identifies passable areas, prohibited areas, and unknown areas in the current three-dimensional spatial map. The collision detection module 603 performs collision detection on each node in the three-dimensional spatial map and constructs a fast exploration random tree. The path planning module 604 plans an initial exploration path for the UAV based on the fast exploration random tree. The path optimization module 605 performs path optimization operations on the initial path to obtain a final path, which the UAV follows for exploration.

[0124] Figure 8 This is a diagram illustrating the internal structure of an electronic device according to one embodiment. The electronic device includes a processor, a memory, a communication interface, and a communication bus. The processor, memory, and communication interface communicate with each other via the communication bus. The memory stores at least one executable instruction, which causes the processor to execute... Figure 1 The process of autonomous exploratory inspection path planning method for Chinese unmanned aerial vehicles (UAVs).

[0125] This application also provides a computer-readable storage medium storing at least one executable instruction. When the executable instruction is executed on an electronic device, it causes the electronic device to perform the operation of the UAV autonomous exploratory inspection path planning method as described in the above embodiments.

[0126] As described above, although the invention has been shown and described with reference to specific preferred embodiments, it should not be construed as limiting the invention itself. Various changes in form and detail may be made without departing from the spirit and scope of the invention as defined in the appended claims.

Claims

1. A method for autonomous exploratory inspection path planning of power unmanned aerial vehicles (UAVs) based on flight safety constraints, characterized in that, Includes the following steps: S1. Acquire real-time environmental information of the power grid equipment area to construct a three-dimensional spatial map; identify all nodes in the three-dimensional spatial map and calculate the probability that each node is an obstacle; S1-1. Construct a three-dimensional spatial map based on an octree. In the three-dimensional spatial map, the time series is recorded as t=1, 2, 3, ..., T, and the information of each time point is recorded as i1, i2, i3, ..., iT. Calculate the probability value of the nth node being occupied. ； S1-2, Based on the probability value of the nth node being occupied. Determine the probability that the nth node is an obstacle: When the probability value This indicates that the current node's state is unknown; when the probability value... This indicates that the current node is idle; when the probability value This indicates that the current node is in an occupied state; S2. Based on the probability of each node being an obstacle calculated in step S1, identify the passable area, the prohibited area, and the unknown area in the current 3D spatial map respectively. S3. Perform collision detection on each node in the three-dimensional spatial map and construct a fast exploration random tree; S3-1. Perform collision detection on each node in the three-dimensional spatial map to find valid sampling points. ; S3-2, Determine the valid sampling points its nearest point If the distance D between the nodes satisfies the threshold condition, output the new node. : ； ； In the formula, Indicates the minimum distance threshold. Indicates the maximum distance threshold; S4. Based on the aforementioned fast exploration random tree, plan the initial exploration path for the UAV; S4-1. Constructing the evaluation function : ； In the formula, The weight value represents the distance from the current node to be selected to the target point; This represents the obstacle avoidance penalty function for the currently selected node; and These represent the distances of the current node from the start point and the end point, respectively. ； ； In the formula, Indicates the coordinates of the starting node. Represented as the coordinates of the currently selected node. Indicates the coordinates of the target point; Obstacle avoidance penalty function The expression is as follows: ； In the formula, For map resolution; Weight parameters of idle nodes; This represents the total number of nodes occupied by candidate node n in the gain calculation model; This represents the total number of nodes occupied by candidate node n in the gain calculation model; S4-2. Using path point n as the center of the gain volume model, traverse the nodes m within the gain model in units of map resolution, determine the node states, and if node m is occupied and observable from path point n, then include it in the gain volume model. If node m is idle and observable with path point n, then it is counted. : ； ； In the formula, This indicates that the current node's state is unknown. This indicates that the current node is in an idle state; This indicates that the current node is in an occupied state; Represents the set of observable nodes; This indicates that m belongs to the set of observable nodes; The gain function representing the occupied node; The gain function representing the idle node; S5. Perform path optimization operation on the initial path to obtain the final path, and the UAV performs exploration according to the final path.

2. The method for autonomous exploratory inspection path planning of power unmanned aerial vehicles according to claim 1, characterized in that, In step S1, environmental information of the power grid equipment area is acquired through a three-dimensional lidar sensor and / or a binocular vision sensor.

3. The autonomous exploratory inspection path planning method for power unmanned aerial vehicles according to claim 1, characterized in that, Step S2 further includes: The current node state is represented by cubes of the same size, indicating passable, impassable, and unknown areas in the map environment, as shown in the following expression: ； In the formula, m represents a node. This indicates that the current node's state is unknown, and the unknown area in the current 3D spatial map is identified based on this node. ; This indicates that the current node is idle, and identifies the passable area in the current 3D spatial map based on this node. ; This indicates that the current node is occupied, and the prohibited area in the current 3D spatial map is identified based on this node. .

4. The method for autonomous exploratory inspection path planning of power unmanned aerial vehicles according to claim 1, characterized in that, Step S5 further includes: S5-1, Simplify the drone to a radius of... The circle, its safe flight distance from obstacles during flight is Therefore, the radius of the safe zone of the path is: ； S5-2. Construct a directed line segment with i as the starting point and i+2 as the ending point. A cube of a predetermined size is constructed with the directed line segment as its center. ; S5-3, Traverse the cube based on map resolution. The state of the nodes within the path is used to determine whether the path can be directly connected. ； In the formula, This indicates that the current node's state is unknown. This indicates that the current node is in an idle state; This indicates that the current node is in an occupied state; like , then represents a line segment If there is an obstacle, retain the path point i+1; like , then represents a line segment There is an unknown region; retain the path point i+1. like , then represents a line segment If all the above states are known and there are no obstacles, then discard path point i+1 and directly connect path point i and path point i+2; S5-4. Repeat steps S5-1 to S5-3 until the end point of the path is found, and the final path is obtained. The drone then performs the exploration according to the final path.

5. A power unmanned aerial vehicle (UAV) autonomous exploratory inspection path planning system, used to execute the power UAV autonomous exploratory inspection path planning method according to any one of claims 1 to 4, characterized in that, include: Map building module; The map building module is used to acquire real-time environmental information of the power grid equipment area and build a three-dimensional spatial map. Identify all nodes in the three-dimensional spatial map and calculate the probability that each node is an obstacle; The region identification module identifies passable areas, prohibited areas, and unknown areas in the current 3D spatial map based on the probability that each node is an obstacle. Collision detection module; the collision detection module is used to perform collision detection on each node in the three-dimensional spatial map and construct a fast exploration random tree; Path planning module; the path planning module plans the initial exploration path of the UAV based on the fast exploration random tree; Path optimization module; The path optimization module is used to perform path optimization operations on the initial path to obtain the final path, and the UAV performs exploration according to the final path.

6. An electronic device, characterized in that, The device includes: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, it implements the autonomous exploratory inspection path planning method for power drones as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The storage medium stores at least one executable instruction, which, when executed on an electronic device, causes the electronic device to perform the autonomous exploratory inspection path planning method for power drones as described in any one of claims 1 to 4.

Images