A method for optimizing UAV airway inspection paths based on dynamic environment perception

Abstract

This invention discloses a method for optimizing UAV airway inspection paths based on dynamic environmental perception, belonging to the field of UAV intelligent path planning technology. The method includes: establishing a three-dimensional environment model based on airway geographic information data and mapping point cloud data, and generating an initial inspection path set with sufficient coverage; using dynamic obstacle trajectories collected by airborne radar and cameras to predict and generate a time-varying obstacle envelope sequence, and performing spatiotemporal overlay calculations with the initial inspection paths to eliminate path segments that conflict with the future movement range of obstacles, resulting in an obstacle avoidance-corrected inspection path set; based on the elevation information in the three-dimensional environment model, predicting the local areas and spatial ranges where communication obstruction may occur in the UAV inspection path, and performing secondary corrections on path segments about to enter the communication obstruction area, outputting the corresponding flight altitude and speed sequences.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned aerial vehicle intelligent path planning, and particularly relates to an unmanned aerial vehicle channel inspection path optimization method based on dynamic environment perception. BACKGROUND

[0002] With the rapid development of unmanned aerial vehicle technology and intelligent perception technology, unmanned aerial vehicles are gradually widely used in the fields of power inspection, traffic monitoring, disaster assessment and logistics distribution. In particular, in the channel inspection scene, unmanned aerial vehicles have the advantages of rapid response, low cost and high mobility, and can perform multi-dimensional inspection in complex geographical environments. However, the traditional channel inspection path planning method mostly relies on static geographic information, such as two-dimensional maps or simplified geographic models, and it is often difficult to fully utilize real-time perception data to model dynamic environments. In recent years, the combination of three-dimensional surveying point cloud technology, laser radar and airborne camera systems enables unmanned aerial vehicles to obtain more detailed spatial structure information during flight and have the ability to dynamically identify obstacles. In addition, researches on the integration of environment modeling and path planning are emerging, trying to assist unmanned aerial vehicles in selecting safer and more efficient routes in the inspection task through three-dimensional environment models.

[0003] However, the existing unmanned aerial vehicle channel inspection path planning has the following deficiencies: on the one hand, although existing technologies can generate three-dimensional environment models by combining geographic information systems and point cloud data, most path planning is still based on static scene assumptions, which is difficult to cope with dynamic obstacles in the channel that change over time, such as construction machinery or temporary floating objects, thereby possibly causing path conflicts and flight risks; on the other hand, existing researches mostly use local correction strategies in path obstacle avoidance, which only circumvents the obstacle position at the current moment, but lack of modeling and predicting the future trajectory of the obstacle, resulting in insufficient global robustness of the planned path; in addition, the communication shadow problem is also a weak link of existing technologies, especially in high bridges, mountain gorges or areas with dense large buildings, signal loss or interruption often occurs in the path of the unmanned aerial vehicle, and existing methods usually do not consider the communication shadow area in the path generation stage, which easily causes interruption of the inspection task or delay of information transmission. SUMMARY

[0004] In view of the problems existing in the prior art unmanned aerial vehicle channel inspection path optimization technology, the present application is proposed.

[0005] Therefore, the problem to be solved by the present application is how to model and correct the dynamic obstacle trajectory and future communication shadow area to improve the safety and continuity of the inspection path.

[0006] To solve the above technical problems, the present application provides the following technical solutions:

[0007] In a first aspect, the present application provides a method for optimizing the inspection path of a UAV in a waterway based on dynamic environment perception, which comprises: establishing a three-dimensional environment model based on waterway geographic information data and surveying point cloud data, and generating an initial inspection path set with a coverage range meeting the requirements; predicting a series of obstacle envelopes changing over time using dynamic obstacle trajectories collected by the onboard radar and camera, and performing spatiotemporal superposition calculation with the initial inspection path to eliminate path segments conflicting with the future movement range of the obstacles, thereby obtaining an inspection path set after obstacle avoidance correction; predicting the local area and spatial range where communication shielding may occur in the UAV inspection path according to the elevation information in the three-dimensional environment model, and performing secondary correction on the inspection path segments about to enter the communication shielding area, and outputting the corresponding flight height and speed sequence.

[0008] As a preferred scheme of the method for optimizing the inspection path of a UAV in a waterway based on dynamic environment perception, the three-dimensional environment model comprises the length of the waterway, the width of the waterway, the water depth topography fluctuation, and the position of the obstacles.

[0009] As a preferred scheme of the method for optimizing the inspection path of a UAV in a waterway based on dynamic environment perception, the generation of the three-dimensional environment model comprises: dividing the waterway surveying point cloud data into a two-dimensional grid according to a fixed spatial resolution Δx, Δy, and recording the average water depth value of each grid point; generating a continuous three-dimensional terrain surface model using triangulation interpolation based on the water depth data of the two-dimensional grid, wherein the spatial range of the three-dimensional terrain surface model corresponds to the length and width of the waterway, and the water depth fluctuation is reflected in the vertical direction; mapping the obstacle detection data into a three-dimensional voxel unit in the waterway coordinate system, and aligning the boundary of the three-dimensional voxel unit with the waterway coordinate system; superimposing and embedding the three-dimensional voxel unit into the three-dimensional terrain surface model to obtain a three-dimensional environment model containing obstacle distribution; assigning an obstacle weight to each three-dimensional voxel unit containing or adjacent to the obstacle; the obstacle weight is calculated by multiplying the reciprocal of the distance between the obstacle and the center point of the three-dimensional voxel unit by the obstacle scale factor.

[0010] As a preferred embodiment of the UAV route inspection path optimization method based on dynamic environment perception described in this invention, the generation of the initial inspection path set includes: projecting a three-dimensional environment model onto a two-dimensional plane, dividing the route into inspection strips along the route width direction, the width of which is the effective scanning bandwidth of the UAV; the two-dimensional plane is the route length × route width; generating a longitudinal main path along the route length direction with the route starting point as the starting point; the longitudinal main path is formed by connecting the left and right boundary points segment by segment and accumulating the three-dimensional voxel units with obstacle weight values ​​higher than a preset weight threshold using a Dijkstra search method between each section; using the longitudinal main path as the skeleton, generating on the left and right sides of each longitudinal path segment sequentially... A lateral scanning path is generated. If a voxel cell with an obstacle weight higher than a preset weight threshold is encountered during the scanning process, the path is locally bypassed and then continues to connect in a serpentine manner. The longitudinal main path and the lateral serpentine scanning path are spliced ​​together to obtain a hybrid path set. For the generated hybrid path set, the coverage and theoretical energy consumption are calculated: the coverage is obtained by statistically analyzing the ratio of the number of three-dimensional voxel cells traversed by the hybrid path to the total number of three-dimensional voxel cells in the channel; the theoretical energy consumption is calculated by weighting the hybrid path length with the obstacle weights of the three-dimensional voxel cells traversed. Hybrid paths with a coverage lower than the coverage threshold or a theoretical energy consumption higher than the energy consumption threshold are removed, and an initial inspection path set is output, while retaining the coverage index and theoretical energy consumption information of each initial inspection path.

[0011] As a preferred embodiment of the UAV airway inspection path optimization method based on dynamic environment perception described in this invention, the method for predicting and generating time-varying obstacle envelope sequences includes: using obstacle trajectory data collected by airborne radar and cameras, including position and velocity vectors, and performing time synchronization and coordinate mapping on the collected obstacle trajectory data to make it consistent with the coordinate system of the three-dimensional environment model; for each dynamic obstacle, using a standard Kalman filter to generate a future position prediction sequence, and generating a three-dimensional envelope of spatial range for the predicted position, and adding a valid time interval to each three-dimensional envelope to form a set of obstacle envelope sequences.

[0012] As a preferred embodiment of the UAV flight path optimization method based on dynamic environment perception described in this invention, the step of eliminating path segments that conflict with the future movement range of obstacles includes: for each inspection path in the initial inspection path set, taking a set of inspection path points, each inspection path point including three-dimensional coordinates; determining the expected arrival time of the inspection path points based on the UAV flight speed and the distance between the inspection path points; checking whether each inspection path point falls within the spatial range of any obstacle envelope in the obstacle envelope sequence, and whether the expected arrival time is within the corresponding time window; if there is a case of falling into the obstacle envelope, then the inspection path point is marked as a conflict point; if there are consecutive conflict points, then the corresponding inspection path segment is removed from the original inspection path, and the inspection path set after obstacle avoidance correction is obtained, and the start and end points of the removed inspection path segments are marked.

[0013] As a preferred embodiment of the UAV airway inspection path optimization method based on dynamic environment perception described in this invention, the method for determining the estimated arrival time of inspection path points includes: assigning an initial takeoff time to the starting point of the inspection path; for any point on the inspection path, the estimated arrival time is obtained by adding the ratio of the spatial distance between the two points to the UAV flight speed based on the estimated arrival time of the previous point; the spatial distance between the two points is calculated by the three-dimensional coordinate difference; the UAV flight speed is a preset constant inspection speed, set according to the UAV model parameters and inspection task requirements.

[0014] As a preferred embodiment of the UAV airway inspection path optimization method based on dynamic environment perception described in this invention, the method for predicting the local area and spatial range where communication obstruction may occur in the UAV inspection path includes: spatially mapping each inspection path in the obstacle avoidance correction set to a three-dimensional environment model to obtain the corresponding terrain height of the inspection path point; for each inspection path point, calculating the elevation difference with the terrain, and if it is less than the set minimum height threshold for UAV communication safety, then marking the inspection path point as potentially having a communication obstruction risk; merging inspection path points with consecutive communication obstruction risks, outputting the inspection path segment with communication obstruction risk, and recording the spatial range and the expected time window of occurrence.

[0015] As a preferred embodiment of the UAV airway inspection path optimization method based on dynamic environment perception described in this invention, the secondary correction of the inspection path segment about to enter the communication obstruction zone includes: superimposing the spatial range of the local area of ​​communication obstruction with the inspection path set after obstacle avoidance correction to determine the inspection path segment about to enter the obstruction zone; recording the start and end positions, time window, and current flight altitude for each inspection path segment; determining the unobstructed area in the three-dimensional environment model whose distance from the inspection path segment about to enter the obstruction zone does not exceed a preset distance threshold as candidate alternative inspection path segments to ensure the continuity of the inspection path and meet the constraints of flight altitude and obstacle avoidance; comprehensively scoring the candidate alternative inspection path segments, and selecting the optimal solution based on a weighted calculation of flight altitude change, obstacle weight, increase in inspection path length, and communication security priority; replacing the inspection path segment about to enter the obstruction zone with the optimized alternative inspection path segment to form a set of inspection paths after communication obstruction correction.

[0016] As a preferred embodiment of the UAV airway inspection path optimization method based on dynamic environment perception described in this invention, the obstacle scale factor is a quantitative parameter calculated based on the spatial size of the obstacle. It is obtained by normalizing the volume of the obstacle in the three-dimensional coordinate system or the projected area in the inspection direction, and is used to characterize the impact of the size of the space occupied by the obstacle on the risk of the local inspection path.

[0017] In a second aspect, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program instructions are executed by the processor, they implement the steps of the UAV route inspection path optimization method based on dynamic environment perception as described in the first aspect of the present invention.

[0018] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program instructions are executed by a processor, the steps of the unmanned aerial vehicle (UAV) airway inspection path optimization method based on dynamic environment perception as described in the first aspect of the present invention are implemented.

[0019] The beneficial effects of this invention are as follows: By integrating airway geographic information data and survey point cloud data to establish a three-dimensional environment model, the inspection path planning becomes comprehensive and fine-grained; by combining obstacle trajectory data collected by airborne radar and cameras, the invention predicts and generates obstacle envelope sequences that change over time, and performs spatiotemporal overlay comparison on the path to eliminate and avoid potential future conflict segments, thereby improving the safety and continuity of the path; in addition, this invention also implements a prediction and secondary correction mechanism for communication obstruction areas, which can dynamically adjust flight altitude and speed in complex terrain or highly obstructed environments to ensure the stability of data transmission.

[0020] Through the above-mentioned multi-level data processing, the present invention can provide more efficient, reliable and intelligent path planning support for UAVs in airway inspection tasks. Attached Figure Description

[0021] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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 effort.

[0022] Figure 1 This is a flowchart of a method for optimizing UAV airway inspection paths based on dynamic environment perception. Detailed Implementation

[0023] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0024] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0025] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0026] Figure 1 This is a flowchart of a UAV airway inspection path optimization method based on dynamic environment perception according to an embodiment of the present invention. Figure 1 As shown, the UAV route inspection path optimization method based on dynamic environment perception includes:

[0027] S1: Based on waterway geographic information data and survey point cloud data, establish a three-dimensional environment model and generate an initial inspection path set with sufficient coverage.

[0028] The three-dimensional environment model includes the channel length, channel width, water depth, topographic relief, and obstacle locations.

[0029] It should be noted that directly planning flight paths on a two-dimensional map cannot accurately reflect the impact of water depth variations and obstacle distribution on UAV flight path selection, leading to potential risks in inspection paths. Therefore, this invention first generates a three-dimensional environment model that corresponds one-to-one with the actual spatial range of the waterway by fusing waterway mapping point cloud data and waterway geographic information data, and then generates a set of candidate initial inspection paths based on this model. This ensures that the inspection path generation process is based on the constraints of a real three-dimensional environment, thereby enhancing the safety and coverage of the paths.

[0030] S1.1: Generate a three-dimensional environment model.

[0031] S1.1.1: Divide the channel survey point cloud data into a two-dimensional grid with fixed spatial resolutions Δx and Δy, and record the average water depth value for each grid point.

[0032] S1.1.2: Based on the water depth data of the two-dimensional grid, a continuous three-dimensional terrain surface model is generated by triangulation interpolation. The spatial range of the three-dimensional terrain surface model corresponds to the channel length and channel width, and reflects the water depth undulation in the vertical direction.

[0033] It should be noted that directly using the grid surface would result in contour lines appearing stepped, making it difficult to accurately represent continuous changes in water depth. This invention connects grid points into an irregular triangular network through triangulation, and uses an interpolation function within the network to estimate the water depth at any point, thereby obtaining a continuous and smooth underwater topographic surface.

[0034] Specifically, firstly, all valid grid points in the two-dimensional grid are triangulated to form a set of non-overlapping triangular units; secondly, a planar interpolation function is established within each triangular unit, and the water depth at any location within the triangular unit is calculated using the water depth value at the vertex of the triangle; finally, all triangular units are spliced ​​together to obtain a continuous three-dimensional terrain surface model.

[0035] S1.1.3: In the channel coordinate system, map the obstacle detection data into three-dimensional voxel units and align the boundaries of the three-dimensional voxel units with the channel coordinate system.

[0036] Obstacle detection data typically originates from sonar scans or lidar point clouds, presenting as irregular point sets or volumetric data. Directly incorporating this data into inspection path planning makes it difficult to establish a unified computational framework. This invention proposes to transform obstacle data into regular three-dimensional voxel units through voxelization, ensuring that the obstacle distribution and the three-dimensional environment model share a consistent coordinate system.

[0037] Specifically, firstly, the voxel side length is defined, and a three-dimensional voxel grid covering the waterway area is constructed in the waterway coordinate system. Then, obstacle detection points are mapped to the corresponding voxels. If the number of obstacle points in a voxel exceeds a set threshold, the corresponding voxel is marked as an obstacle voxel. In this way, obstacle data is uniformly encoded into voxel units compatible with the terrain model. Finally, the boundaries of all voxel units are aligned with the X, Y, and Z axes of the waterway coordinate system, thus ensuring ease of indexing and calculation in the subsequent path planning process.

[0038] S1.1.4: Three-dimensional voxel units are superimposed and embedded into a three-dimensional terrain surface model to obtain a three-dimensional environment model containing obstacle distribution.

[0039] S1.1.5: Assign obstacle weights to each 3D voxel unit that contains or is adjacent to an obstacle.

[0040] Conventional methods typically employ binarization, marking obstacle voxels as "impassable" and non-obstacle voxels as "passable." This approach fails to reflect the potential risks in the obstacle's neighborhood, easily leading to planned inspection paths running too close to obstacle boundaries, creating collision hazards. To address this, this invention proposes calculating obstacle weights based on a combination of obstacle distance and scale factors, enabling inspection path planning to perceive the obstacle's influence range and size.

[0041] The specific calculation method is as follows: the obstacle weight is calculated by multiplying the inverse of the distance between the obstacle and the center point of the three-dimensional voxel unit with the obstacle scale factor, so as to reflect the comprehensive impact of the obstacle's position and volume on local spatial risk.

[0042] The distance between the obstacle and the center point of the three-dimensional voxel unit can be calculated using Euclidean distance.

[0043] Preferably, the obstacle scale factor is a quantitative parameter calculated based on the spatial dimensions of the obstacle. It is obtained by normalizing the volume of the obstacle in the three-dimensional coordinate system or the projected area in the inspection direction, and is used to characterize the impact of the size of the space occupied by the obstacle on the risk of the local inspection path.

[0044] Using the above calculation method, the closer an obstacle is to the center point of a voxel, or the larger its spatial scale, the higher its obstacle weight value, thereby effectively improving the ability to avoid high-risk areas during the inspection path planning process.

[0045] S1.2: Generate the initial inspection path set.

[0046] S1.2.1: Project the three-dimensional environment model onto a two-dimensional plane and divide it into inspection strips along the width of the channel. The width of the inspection strip is the effective scanning bandwidth of the UAV. The two-dimensional plane is the channel length × channel width.

[0047] S1.2.2: Starting from the channel starting point, a longitudinal main path is generated along the channel length direction; the longitudinal main path is formed by connecting the left and right boundary points segment by segment and accumulating them by using a Dijkstra search-based method between each section to avoid three-dimensional voxel units with obstacle weight values ​​higher than the preset weight threshold.

[0048] It should be noted that when generating the inspection path, a longitudinal master path must first be established as the skeleton for the subsequent lateral scan path generation. Conventional methods often involve generating a straight trajectory directly along the channel centerline, but this approach cannot guarantee safety in the presence of obstacles or significant water depth fluctuations. This invention proposes a longitudinal path generation method based on Dijkstra's search, which can form a more robust longitudinal master path while considering obstacle weights.

[0049] In this embodiment of the invention, the specific operations are as follows:

[0050] First, the channel is divided into multiple sections along its length, each section consisting of transverse grid points. Second, the left and right boundary points on each section are selected as candidate nodes, and the connectivity cost between nodes in adjacent sections is calculated. The connectivity cost is obtained by weighting distance and obstacle weights. Third, Dijkstra's search algorithm is used to expand the channel from the starting point downstream section by section, selecting the node connection sequence with the minimum cumulative connectivity cost. Finally, the trajectories of all connected nodes are spliced ​​together to form the longitudinal main path.

[0051] In this way, the longitudinal main path can not only run through the entire length of the waterway, but also effectively avoid dangerous areas where the obstacle weight is higher than the threshold.

[0052] S1.2.3: Using the longitudinal main path as the skeleton, transverse scanning paths are generated sequentially on the left and right sides of each longitudinal path segment. If a voxel unit with an obstacle weight higher than the preset weight threshold is encountered during the scanning process, the serpentine connection continues after local bypass.

[0053] It should be noted that after generating the longitudinal main path, to ensure that the inspection coverage meets the full width of the waterway, lateral scan paths need to be generated on both sides of the longitudinal path. The lateral scan paths are formed by sequentially arranging scan lines along the left and right sides of each segment of the longitudinal path to cover the areas not covered by the longitudinal path, thus creating a serpentine scan trajectory.

[0054] In this embodiment of the invention, the specific operations include:

[0055] First, an initial set of lateral path points is generated at the start and end points of each path segment along the width of the airway. The spacing between the path points is set according to the effective scanning bandwidth of the UAV, so that the scanning path can cover the left and right sides of the longitudinal path. Then, adjacent path points are connected in sequence to form lateral scanning path segments, forming a continuous serpentine trajectory.

[0056] During the connection process, if the obstacle weight of the corresponding 3D voxel unit is found to be higher than the preset threshold for any path point, local bypass is performed at the path point. That is, under the condition of not exceeding the maximum offset distance, the position of the path point is adjusted to bypass the obstacle while maintaining the continuity of the path segment.

[0057] S1.2.4: Combine the vertical main path and the horizontal serpentine scan path to obtain a mixed path set.

[0058] First, the longitudinal main path is segmented, with the length of each segment matching the strip length of the transverse scanning path. Second, transverse paths are inserted on both sides of each segment in a serpentine pattern, ensuring that the start and end points of the transverse paths are precisely connected to the nodes of the longitudinal main path. Third, a path smoothing algorithm (such as B-spline interpolation) is used to locally adjust the splicing points to avoid abrupt path turns that could lead to flight instability. Finally, a set of interwoven longitudinal and transverse paths is obtained.

[0059] The paths generated in this way ensure both the continuity of inspections and the coverage of the waterway area, providing a variety of candidate solutions for subsequent path optimization.

[0060] It should be noted that while using only the vertical main path ensures path continuity, it results in insufficient coverage in the horizontal direction. Conversely, using only the horizontal serpentine scanning path provides high coverage but is lengthy and lacks global guidance. Therefore, this invention employs a path splicing operation, using the vertical main path as the overall framework and embedding horizontal serpentine paths into each vertical path segment to achieve a combination of vertical and horizontal coverage.

[0061] S1.2.5: Calculate coverage and theoretical energy consumption for the generated mixed path set:

[0062] Coverage is obtained by statistically analyzing the ratio of the number of three-dimensional voxel units traversed by the mixed path to the total number of three-dimensional voxel units in the channel; theoretical energy consumption is calculated by weighting the mixed path length with the obstacle weights of the three-dimensional voxel units traversed.

[0063] S1.2.6: Remove mixed paths with coverage rates below the coverage threshold or theoretical energy consumption above the energy consumption threshold, output the initial inspection path set, and retain the coverage index and theoretical energy consumption information of each initial inspection path.

[0064] Through the above filtering operations, the final initial inspection path set not only has high coverage efficiency, but also ensures feasibility under energy consumption constraints, thus providing reliable inspection path candidates for actual UAV route inspection.

[0065] S2: Using the dynamic obstacle trajectories collected by airborne radar and cameras, predict and generate a time-varying obstacle envelope sequence, and perform spatiotemporal superposition calculation with the initial inspection path to remove path segments that conflict with the future movement range of obstacles, thus obtaining the inspection path set after obstacle avoidance correction.

[0066] S2.1: Predict and generate a time-varying sequence of obstacle envelopes.

[0067] Obstacle trajectory data, including position and velocity vectors, is collected using airborne radar and cameras. The collected obstacle trajectory data is time-synchronized and coordinate-mapped to be consistent with the coordinate system of the three-dimensional environment model. For each dynamic obstacle, a future position prediction sequence is generated using a standard Kalman filter. A three-dimensional envelope of the spatial range is generated for the predicted position, and a valid time interval is added to each three-dimensional envelope to form a set of obstacle envelope sequences.

[0068] S2.2: Eliminate path segments that conflict with the future movement range of obstacles.

[0069] After obtaining the obstacle envelope sequence set, it is necessary to compare it point by point with the initial inspection path set to identify and eliminate conflicting path segments, thereby obtaining the path set after obstacle avoidance correction.

[0070] The above process not only considers whether the path point coincides with the obstacle location, but also maps the expected arrival time of the path point to the time window of the obstacle envelope to ensure spatiotemporal consistency.

[0071] S2.2.1: For each inspection path in the initial inspection path set, take the inspection path point set, where each inspection path point includes three-dimensional coordinates, namely horizontal position, vertical position and height information.

[0072] This discretization method transforms the path from a continuous curve into a finite sequence of discrete points, facilitating subsequent point-by-point checks to determine whether it overlaps with the obstacle envelope.

[0073] S2.2.2: Determine the estimated arrival time of the inspection path points based on the UAV's flight speed and the distance between inspection path points.

[0074] Determining the estimated arrival time of inspection path points includes: assigning an initial takeoff time to the starting point of the inspection path; for any point on the inspection path, the estimated arrival time is obtained by adding the ratio of the spatial distance between the two points to the UAV's flight speed to the estimated arrival time of the previous point.

[0075] The spatial distance between two points is calculated by the difference in three-dimensional coordinates.

[0076] The drone's flight speed is a preset constant inspection speed, set according to the drone model parameters and inspection task requirements. This constant speed assumption simplifies the time calculation process and avoids prediction uncertainties caused by real-time speed fluctuations.

[0077] S2.2.3: For each inspection path point, check whether it falls within the spatial range of any obstacle envelope in the obstacle envelope sequence, and whether the expected arrival time is within the time window corresponding to the obstacle envelope. If both spatial and temporal conditions are met, it indicates that the UAV will enter the future location area of ​​the obstacle at the corresponding time point, i.e., there is a situation where it falls into the obstacle envelope, thus constituting a path conflict. In this case, the corresponding inspection path point should be marked as a conflict point.

[0078] If there are consecutive conflict points (e.g., at least 3 consecutive conflict points), the corresponding inspection path segment will be removed from the original inspection path. The inspection path set after obstacle avoidance correction will be generated, and the start and end points of the removed inspection path segments will be marked.

[0079] It can be seen that the inspection path set after obstacle avoidance correction avoids conflict with the future movement range of dynamic obstacles, and improves the feasibility and safety of the inspection path.

[0080] S3: Based on the elevation information in the 3D environment model, predict the local areas and spatial ranges where communication may be blocked in the UAV inspection path, perform secondary correction on the inspection path segment that is about to enter the communication blockage area, and output the corresponding flight altitude and speed sequence.

[0081] S3.1: Predict the local areas and spatial ranges where communication may be obstructed during the UAV inspection path.

[0082] First, each inspection path in the set of inspection paths after obstacle avoidance correction is spatially mapped to the three-dimensional environment model to obtain the corresponding terrain height of the inspection path point.

[0083] The mapping operation involves spatially matching the horizontal and vertical coordinates of each inspection path point with the grid or continuous surface of the 3D environment model, and obtaining the elevation values ​​of the corresponding grid points or interpolated surfaces, which facilitates the calculation of the relative height difference between the UAV and ground obstacles.

[0084] Through the above mapping, each inspection path point not only has its own spatial coordinates, but also corresponding terrain elevation information, providing an accurate basis for determining communication obstruction.

[0085] Secondly, for each inspection path point, the elevation difference with the terrain is calculated. If it is less than the set minimum height threshold for drone communication safety, it indicates that the drone may be in an area where communication signals are blocked, and the inspection path point is marked as potentially having a communication obstruction risk.

[0086] The minimum altitude threshold for drone communication security can be set autonomously based on the characteristics of the drone communication link, antenna gain, signal attenuation parameters, and safety requirements of the inspection mission, so as to ensure that the drone can maintain basic communication quality near the marked point.

[0087] The inspection path points with continuous communication blockage risk are merged, and the inspection path segments with communication blockage risk are output, and the spatial range and the expected time window are recorded.

[0088] S3.2: Perform secondary correction on the inspection path segment that is about to enter the communication obstruction area.

[0089] S3.2.1: Overlay the spatial range of the communication obstruction area with the inspection path set after obstacle avoidance correction to determine the inspection path segment that is about to enter the obstruction area, and record the start and end positions, time window and current flight altitude for each inspection path segment.

[0090] The overlay operation includes aligning the path segment with the original path in three-dimensional space and combining it with time window information to determine the specific inspection path segment that the UAV will enter during the inspection process.

[0091] During the recording process, each inspection path segment includes the start point, end point, expected time window, and current flight altitude, providing precise spatial and temporal constraints for subsequent alternative path planning.

[0092] S3.2.2: Uncovered areas in the 3D environment model whose distance from the inspection path segment about to enter the covert area does not exceed a preset distance threshold are identified as candidate alternative inspection path segments to ensure the continuity of the inspection path and meet the flight altitude and obstacle avoidance constraints.

[0093] These candidate alternative inspection route segments are spatially adjacent to the shielded area, which can ensure the continuity of the route while meeting flight altitude constraints and obstacle avoidance requirements.

[0094] S3.2.3: A comprehensive score is given to the candidate alternative inspection path segments, and the optimal solution is selected based on a weighted calculation of changes in flight altitude, obstacle weight, increase in inspection path length, and communication security priority.

[0095] The path segment about to enter the obstruction zone is replaced with an optimized alternative path segment, forming a set of inspection paths corrected for communication obstruction. This updated path set not only considers obstacle avoidance and communication safety but also outputs the flight altitude and speed sequence for each path, providing complete execution data for subsequent UAV inspection tasks. Through these operations, the UAV can dynamically adapt to communication obstruction risks in complex flight path environments, achieving safe and continuous inspections.

[0096] For each inspection path segment, the drone's flight altitude, terrain elevation, and spatial distance to surrounding obstacles are acquired along the path. By comparing the difference between the drone's flight altitude and the terrain elevation, if the difference is close to or below the minimum communication security altitude threshold, the communication security risk of the path segment increases, and the corresponding communication security priority is set to a higher value. Conversely, if the flight altitude is significantly higher than the surrounding terrain and obstacles are sparse, the communication security risk is low, and the communication security priority is set to a lower value. Specific settings can be adjusted as needed.

[0097] This embodiment also provides a computer device applicable to the UAV route inspection path optimization method based on dynamic environment perception, including a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the UAV route inspection path optimization method based on dynamic environment perception as proposed in the above embodiment.

[0098] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0099] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the method for optimizing UAV airway inspection paths based on dynamic environment perception as proposed in the above embodiments.

[0100] In summary, this invention establishes a three-dimensional environment model by integrating airway geographic information data and survey point cloud data, enabling comprehensive and fine-grained inspection path planning. By combining obstacle trajectory data collected by airborne radar and cameras, it predicts and generates a time-varying obstacle envelope sequence and performs spatiotemporal overlay comparison of the path, enabling the elimination and avoidance of potential future conflict segments, thereby improving path safety and continuity. Furthermore, this invention also incorporates a prediction and secondary correction mechanism for communication obstruction areas, dynamically adjusting flight altitude and speed in complex terrain or highly obstructed environments to ensure data transmission stability.

[0101] Through the above-mentioned multi-level data processing, the present invention can provide more efficient, reliable and intelligent path planning support for UAVs in airway inspection tasks.

[0102] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for optimizing UAV airway inspection paths based on dynamic environment perception, characterized in that: The application relates to a method for generating an initial inspection path set of a UAV (unmanned aerial vehicle) in a waterway, and belongs to the technical field of UAV inspection. The method comprises the following steps: based on waterway geographic information data and surveying point cloud data, a three-dimensional environment model is established, and an initial inspection path set covering a range meeting requirements is generated; By using dynamic obstacle trajectories collected by an airborne radar and a camera, a sequence of time-varying obstacle envelopes is predicted and generated, and spatiotemporal superposition calculation is performed on the initial inspection path, so that path segments in conflict with the future obstacle movement range are removed, and an inspection path set after obstacle avoidance correction is obtained; According to the elevation information in the three-dimensional environment model, local areas and spatial ranges where communication shielding may occur in the UAV inspection path are predicted, the inspection path segments about to enter the communication shielding area are secondarily corrected, and corresponding flight height and speed sequences are outputted. The generation of the three-dimensional environment model comprises the following steps: waterway surveying point cloud data is divided into a two-dimensional grid according to fixed spatial resolutions Dx and Dy, and an average water depth value is recorded for each grid point; Based on the water depth data of the two-dimensional grid, a continuous three-dimensional terrain surface model is generated by using triangulation interpolation, the spatial range of the three-dimensional terrain surface model corresponds to the length and width of the waterway, and the water depth fluctuation is reflected in the vertical direction; in the waterway coordinate system, obstacle detection data is mapped into three-dimensional voxel units, and the boundaries of the three-dimensional voxel units are aligned with the waterway coordinate system; the three-dimensional voxel units are superimposed and embedded into the three-dimensional terrain surface model, and a three-dimensional environment model containing obstacle distribution is obtained; an obstacle weight is given to each three-dimensional voxel unit containing or adjacent to an obstacle; the obstacle weight is calculated by multiplying the reciprocal of the distance between the obstacle and the center point of the three-dimensional voxel unit by an obstacle scale factor. 2.The dynamic environment perception based UAV route inspection path optimization method of claim 1, wherein: The three-dimensional environment model comprises the length, width, water depth terrain fluctuation and obstacle position of the waterway. 3.The method of claim 2, wherein: The generation of the initial inspection path set comprises the following steps: The three-dimensional environment model is projected onto a two-dimensional plane, and inspection strips are divided according to the width direction of the waterway, and the inspection strip width is the effective scanning bandwidth of the UAV; the two-dimensional plane is the length of the waterway multiplied by the width of the waterway; A longitudinal main path is generated along the length direction of the waterway with the starting point of the waterway as the starting point; the longitudinal main path is connected to the left and right boundary points and accumulated to form by avoiding the three-dimensional voxel units with an obstacle weight higher than a preset weight threshold value in each section through Dijkstra search based obstacle avoidance; Based on the longitudinal main path as a skeleton, transverse scanning paths are generated on the left and right sides of each longitudinal path segment in sequence; if a voxel unit with an obstacle weight higher than the preset weight threshold value is encountered in the scanning process, the scanning path is continued to be connected in a serpentine manner after local obstacle avoidance; The longitudinal main path and the transverse serpentine scanning path are spliced to obtain a mixed path set; The coverage rate and the theoretical energy consumption of the generated mixed path set are calculated: The coverage rate is obtained by calculating the ratio of the number of three-dimensional voxel units passed by the mixed path to the total number of three-dimensional voxel units of the waterway; The theoretical energy consumption is calculated by adding the length of the mixed path and the weighted sum of the obstacle weights of the three-dimensional voxel units passed by the mixed path; The mixed paths with a coverage rate lower than a coverage rate threshold value or a theoretical energy consumption higher than an energy consumption threshold value are removed, and the initial inspection path set is outputted, and the coverage rate index and the theoretical energy consumption information of each initial inspection path are retained. 4.The method of claim 3, wherein: The prediction generates a sequence of obstacle envelopes changing over time, comprising: Collecting obstacle trajectory data including position and velocity vector by onboard radar and camera, and time synchronizing and coordinate mapping the collected obstacle trajectory data to be consistent with the coordinate system of the three-dimensional environment model; For each dynamic obstacle, generating a sequence of future position predictions by a standard Kalman filter, and generating a three-dimensional envelope body of spatial range for the predicted position, and attaching a time valid interval to each three-dimensional envelope body to form a set of obstacle envelope sequences. 5.The dynamic environment-aware based UAV route inspection path optimization method of claim 4, wherein: The removing of path segments conflicting with the future obstacle movement range comprises: For each inspection path in the initial set of inspection paths, taking a set of inspection path points, each including a three-dimensional coordinate; Determining the expected arrival time of an inspection path point according to the flight speed of the UAV and the distance between inspection path points; Checking whether each inspection path point falls within the spatial range of any obstacle envelope in the obstacle envelope sequence and whether the expected arrival time is within the corresponding time window; If there is a case of falling into an obstacle envelope, marking the inspection path point as a conflict point; If there are consecutive conflict points, removing the corresponding inspection path segment from the original inspection path, and marking the start and end points of the removed inspection path segment. 6.The dynamic environment-aware based UAV route inspection path optimization method of claim 5, wherein: The determination of the expected arrival time of an inspection path point comprises: Assigning the initial takeoff time to the start point of the inspection path; For any point on the inspection path, the expected arrival time is obtained by adding the ratio of the spatial distance between the two points and the flight speed of the UAV to the expected arrival time of the previous point; The spatial distance between the two points is calculated by the difference of three-dimensional coordinates; The flight speed of the UAV is a preset constant inspection speed set according to the UAV model parameters and the inspection task requirements. 7.The dynamic environment-aware based UAV route inspection path optimization method of claim 6, wherein: The prediction of local areas and spatial ranges where communication shielding may occur in the UAV inspection path comprises: Spatially mapping each inspection path in the set of inspection paths after obstacle avoidance correction with the three-dimensional environment model to obtain the corresponding terrain height of the inspection path point; For each inspection path point, calculating the terrain elevation difference, and if it is less than the set minimum height threshold of the UAV communication safety, marking the inspection path point as possibly having a communication shielding risk; Merging consecutive inspection path points with communication shielding risks to output inspection path segments with communication shielding risks, and recording the spatial range and the time window when it is expected to occur. 8.The dynamic environment-aware based UAV route inspection path optimization method of claim 7, wherein: The secondary correction of the inspection path segment about to enter the communication shielding area comprises: Superimposing the spatial range of the communication shielding local area on the set of inspection paths after obstacle avoidance correction to determine the inspection path segment about to enter the shielding area, and recording the start and end positions, time window, and current flight height for each inspection path segment; Determining the unshielded area in the three-dimensional environment model that is within a preset distance threshold from the inspection path segment about to enter the shielding area as a candidate alternative inspection path segment, ensuring the continuity of the inspection path and meeting the flight height and obstacle avoidance constraints; The candidate alternative inspection path segments are comprehensively scored, and an optimal scheme is selected according to the weighted calculation of the flight height change, the obstacle weight, the length increase of the inspection path and the communication safety priority; The inspection path segment about to enter the shielding area is replaced by the optimized alternative inspection path segment, and a communication shielding corrected inspection path set is formed. 9.The dynamic environment-aware based UAV route inspection path optimization method of claim 8, wherein: The obstacle scale factor is a quantitative parameter calculated according to the spatial size of the obstacle, which is obtained by normalizing the volume of the obstacle in the three-dimensional coordinate system or the projection area of the obstacle in the inspection direction, and is used to represent the influence of the size of the obstacle occupying space on the risk of the local inspection path.

Images