Unmanned aerial vehicle intelligent inspection method and system combining three-dimensional point cloud modeling and route planning technology

Abstract

The invention provides an unmanned aerial vehicle intelligent inspection method and system combining a three-dimensional point cloud modeling and flight path planning technology. The system comprises a data acquisition module, a data processing module, a platform uploading module, a three-dimensional point cloud modeling module, a flight path planning module and a task execution module. By combining the three-dimensional point cloud modeling and flight path planning technology, the whole process is automatic, a closed loop is formed from data acquisition to flight path generation, manual intervention is reduced, automation, refinement and intelligentization of unmanned aerial vehicle routing inspection are achieved, and the routing inspection efficiency and safety are remarkably improved. The method has the advantages that point cloud data is processed through a pre-constructed point cloud data processing model, the quality of the point cloud data is improved, sub-centimeter-level modeling is realized in combination with a BIM technology, the precision of three-dimensional point cloud modeling is high, track planning is based on real three-dimensional geometry, and fine inspection is supported. And through data storage of different file types, the transmission processing efficiency is improved, the space information of the three-dimensional point cloud is fully utilized in the route planning, and refined inspection is supported.

Description

Technical Field

[0001] The present application relates to the field of drone intelligent inspection technology, and in particular to a drone intelligent inspection method and system that combines three-dimensional point cloud modeling and trajectory planning technology. Background Art

[0002] In recent years, with the rapid development of unmanned aerial vehicle (UAV) technology, its application in the power industry has become increasingly widespread, especially in the field of power transmission channel inspection. Compared with traditional manual inspection methods, drone inspections offer the advantages of high efficiency, safety, and low cost. They can effectively reduce the safety hazards associated with manual climbing of towers and traversing complex terrain, and significantly improve inspection efficiency. Currently, mainstream transmission drone inspection methods mainly rely on visible light / infrared cameras to capture images or videos, transmit the collected data to artificial intelligence (AI) platforms for analysis, and use deep learning models (such as target detection and semantic segmentation) to detect transmission equipment faults and identify defects (such as broken insulators, broken conductors, and missing bolts).

[0003] However, this approach still faces the following key challenges in practical applications: 1. The accuracy of 3D point cloud modeling is insufficient, making it difficult to support detailed inspections. Current drone inspections primarily rely on 2D images or low-precision point clouds for defect identification, while transmission channel environments are complex, involving multiple targets such as conductors, towers, insulators, hardware, and surrounding buildings and trees. Existing methods have the following limitations in 3D reconstruction: Large point cloud registration errors: Traditional ICP (Iterative Closest Point) algorithms are susceptible to cumulative errors in long-distance transmission line scenarios, resulting in misalignment and distortion in the stitched 3D model. In particular, the geometric structure of large-scale targets such as buildings is poorly restored. Lack of semantic information: Most point cloud modeling methods fail to incorporate semantic segmentation, making it difficult to distinguish power equipment from the background (such as vegetation and buildings), and failing to provide effective environmental awareness for trajectory planning. Poor adaptability to dynamic scenarios: Existing methods are insufficiently capable of modeling dynamic obstacles such as temporary construction equipment and vehicles, impacting the real-time and safety of drone obstacle avoidance. 2. Lack of an efficient model management platform. Existing inspection systems have significant shortcomings in data and model management: Data silos: Massive amounts of image and point cloud data generated by inspections are stored in disparate locations, lacking unified standards and correlation mechanisms, making it difficult to support comparative analysis of historical data. Model update lags: Defect detection models typically use static training sets and are unable to adapt to seasonal and regional changes in equipment status (e.g., false detections caused by snow-covered insulators). Weak visualization capabilities: Most platforms only provide 2D reports or simple 3D displays, failing to achieve digital twin-level interactive analysis, hindering efficient operations and maintenance decision-making. 3. Route planning is disconnected from the 3D environment. Current drone inspection routes often rely on preset routes or manual remote control, lacking intelligence. Insufficient environmental perception: Route planning fails to fully utilize the spatial information of 3D point clouds, making it difficult to avoid new obstacles (e.g., illegal buildings and overly tall trees). Low mission flexibility: Fixed routes cannot dynamically adjust inspection priorities based on equipment defect risk levels (e.g., requiring additional camera angles in high-fault areas). Lack of energy efficiency optimization: The impact of environmental factors such as wind speed and sunlight on flight stability and battery life is not considered, resulting in ineffective flights or redundant data.

[0004] With the construction of ultra-high voltage (UHV) power grids and the increasing density of urban transmission corridors, traditional inspection methods are no longer able to meet the demands of high-precision, fully automated, and intelligent O&M. Therefore, designing an intelligent drone inspection method that integrates high-precision 3D point cloud modeling with adaptive trajectory planning, and building a standardized model management platform, is a key breakthrough in improving the quality and efficiency of power transmission O&M. Summary of the Invention

[0005] In view of this, it is necessary to provide a drone intelligent inspection method that combines 3D point cloud modeling and trajectory planning technology. By combining 3D point cloud modeling and trajectory planning technology, the entire process is automated, forming a closed loop from data collection to trajectory generation, reducing manual intervention, and realizing the automation, refinement and intelligence of drone inspections, significantly improving inspection efficiency and safety.

[0006] The present application provides a method and system for intelligent inspection of unmanned aerial vehicles (UAVs) that combines three-dimensional point cloud modeling and trajectory planning technology, including: data collection, data processing, uploading to a platform, three-dimensional point cloud modeling, trajectory planning, and task execution modules. By combining three-dimensional point cloud modeling and trajectory planning technology, the entire process is automated, forming a closed loop from data collection to trajectory generation, reducing manual intervention, and realizing the automation, refinement, and intelligence of UAV inspections, significantly improving inspection efficiency and safety. High-precision modeling and planning, processing point cloud data through a pre-built point cloud data processing model to improve the quality of point cloud data, combining BIM technology to achieve sub-centimeter-level modeling, three-dimensional point cloud modeling with high accuracy, trajectory planning based on real three-dimensional geometry, and supporting refined inspections. By saving data in different file types, the transmission processing efficiency is improved, and route planning makes full use of the spatial information of three-dimensional point clouds to support refined inspections.

[0007] In a first aspect, an embodiment of the present application provides a drone intelligent inspection method that combines three-dimensional point cloud modeling and trajectory planning technology, the method comprising:

[0008] S1: Data collection: Scan the target detection points using the 3D laser scanning device on the drone, perform point cloud data stitching, and save it as a point cloud dataset in a LAS file;

[0009] S2: Data processing: Perform Autodesk ReCap cutting on the point cloud dataset, convert the format, and save it as a point cloud dataset of rcs file;

[0010] S3: Upload to the platform: upload the processed data set to the UAV flight control system, and the UAV flight control system supplements and fits the spatial point cloud data through spatial interpolation method;

[0011] S4: 3D point cloud modeling: The UAV flight control system performs BIM modeling on the point cloud dataset through modeling technology, saves it as an rvt file, and imports it into a visualization platform for lightweight display;

[0012] S5: Trajectory planning: The UAV flight control system plans a trajectory covering all detection points based on the constructed point cloud model, generates a trajectory planning file, and sends it to the UAV;

[0013] S6: Mission execution: The drone receives the trajectory planning file and performs the inspection mission.

[0014] Optionally, in an implementation of the first aspect of the present invention, the step S1: data acquisition: scanning target detection points using a three-dimensional laser scanning device on a drone, performing point cloud data stitching, and saving the point cloud data set as a LAS file includes:

[0015] Use pre-built point cloud data processing models to filter and stitch laser point cloud data;

[0016] The point cloud data processing model is constructed by the following steps:

[0017] S1.1: Divide the target point cloud Q into a voxel grid and calculate the normal distribution of points within each voxel; optimize the probability scoring function to maximize the overlap between the projection of the source point cloud P and the distribution of Q;

[0018] S1.2: Use the corner detection algorithm to extract key points with obvious features by calculating the curvature change of the point cloud, and output the simplified point cloud P o and Q o ;

[0019] S1.3: P o For each point in Q o Find the points with the closest Euclidean distance as the corresponding point pair, solve the optimal R and T through SVD decomposition, and minimize the mean square error (MSE) of the corresponding point pairs. Stop when the average distance d is less than the threshold or reaches the maximum number of iterations.

[0020]

[0021] Among them, p i ∈P o ,q i ∈Q o ,i,n is the number of point cloud data sets, R is the matrix obtained by rotation change, and T is the matrix obtained by translation change;

[0022] S1.4: Randomly sample multiple candidate point pairs, use least squares to fit R and T, and select the model with the most point pairs whose statistical error is less than the threshold as the final registration result as the initial estimate.

[0023] S1.5: Adjust the point position by iterative nonlinear least squares method, and update the point position by iterative matrix J and matrix H, where

[0024]

[0025] H=J T J,

[0026]

[0027] p k+1 =p k +αΔp,

[0028] Among them, α represents the iterative step size of each update p, represents the mean value of the error;

[0029] S1.6: After the iteration, the point cloud data is stitched and saved as a point cloud dataset in a las file.

[0030] Optionally, in an implementation of the first aspect of the present invention, the step S2: data processing: performing Autodesk ReCap segmentation on the point cloud dataset, performing format conversion, and saving the point cloud dataset as an rcs file includes:

[0031] Import point cloud data, open Autodesk ReCap software, and select "Open" from the "File" menu to import the LAS point cloud data file to be processed;

[0032] Cutting point cloud data: In ReCap, the user selects a specific area through the Limit Box to cut the point cloud data;

[0033] Clean and optimize data. After cutting is completed, use the cleaning tools provided by ReCap to remove useless data.

[0034] Convert to RCS format, and after completing the cutting and cleaning, export the RCS file and save it.

[0035] Optionally, in an implementation of the first aspect of the present invention, the step S3: uploading to the platform: uploading the processed data set to a UAV flight control system, wherein the UAV flight control system supplements and fits the spatial point cloud data using a spatial interpolation method, including:

[0036] Preprocessing of raw point cloud data, including filtering, denoising, outlier detection, and data correction;

[0037] The improved inverse distance weighted method is used for spatial interpolation, and the formula is:

[0038]

[0039] Among them, let Z i is the attribute value at the reference point, u represents the index, d i Represents the Euclidean distance between the interpolation point p and the reference point. According to the characteristics of the interpolation function, when point p is infinitely close to the reference point, d i→0, the function value f1(p) will converge to Z i ; In particular, when d i = 0, f1(p) = Z i ;

[0040] The processed point cloud data is transmitted to the UAV flight control system via wireless signals. After receiving the data, the system uses spatial interpolation to supplement and fit the point cloud, and converts the discrete point cloud into a continuous surface through triangulation or rasterization technology. The ground point cloud data collected by the UAV is used to generate DEM and DSM, and the model is verified and evaluated for accuracy. It is combined with other sensor data for time alignment and spatial registration.

[0041] Optionally, in an implementation of the first aspect of the present invention, the S4: 3D point cloud modeling: the UAV flight control system performs BIM modeling on the point cloud dataset using modeling technology, saves it as an rvt file, and imports it into a visualization platform for lightweight display, including:

[0042] Use 3D modeling software to import the processed point cloud data and generate a preliminary 3D model, perform triangulation and texture mapping, and convert the discrete point cloud data into a continuous 3D surface model;

[0043] Based on the 3D model, the model is optimized through BIM modeling software, the attributes and parameters required by the Building Information Model (BIM) are added, the point cloud data is combined with the historical image data for analysis, deformation feature points are extracted, and a detailed BIM model is generated;

[0044] The final generated model is saved in .rvt format.

[0045] Optionally, in an implementation of the first aspect of the present invention, S5: trajectory planning: the UAV flight control system performs trajectory planning covering all detection points based on the constructed point cloud model, generates a trajectory planning file, and sends it to the UAV, including:

[0046] Through the 3D point cloud, the spatial information of key target points is extracted to generate a preliminary route;

[0047] According to the inspection objectives, parameters are designed for each point of interest, including shooting angle, distance, and height, and waypoints are planned in sequence according to the order of parameters;

[0048] Use Dijkstra algorithm to optimize path planning;

[0049] Add auxiliary points and set a reasonable safety distance between adjacent waypoints to prevent the drone from accidentally hitting the equipment during flight and ensure the safety of the route;

[0050] The generated route file should be stored in KML format for easy subsequent use.

[0051] Optionally, in an implementation of the first aspect of the present invention, extracting spatial information of key target points through a three-dimensional point cloud to generate a preliminary route includes:

[0052] Extract the spatial information of the target point through 3D point cloud data, and generate a preliminary route using route planning software;

[0053] Before said generated route file should be stored in KML format include:

[0054] According to inspection requirements, set the shooting angle, pitch angle, shooting distance and hovering time parameters for each point of interest;

[0055] Arrive at waypoints in the predetermined order and perform photo-taking tasks to ensure that all inspection targets are covered;

[0056] After the route planning is completed, the route is rehearsed in a simulated environment to check for safety hazards or missed areas. Waypoints and photography parameters are adjusted through real-time simulation to ensure the safety and integrity of the route.

[0057] Manual review and fine-tuning, based on actual on-site conditions, fine-tune the waypoint positions and parameters to ensure that the route meets actual needs.

[0058] In a second aspect, an embodiment of the present application provides a drone intelligent inspection system that combines three-dimensional point cloud modeling and trajectory planning technology, which is applied to the drone intelligent inspection method that combines three-dimensional point cloud modeling and trajectory planning technology as described in the first aspect, and is characterized by including:

[0059] The data acquisition module is used to scan the target detection points through the 3D laser scanning device on the UAV, stitch the point cloud data, and save it as a point cloud dataset in the form of a LAS file;

[0060] A data processing module is used to perform Autodesk ReCap cutting on the point cloud dataset, convert the format of the point cloud dataset, and save it as an rcs file;

[0061] A data transmission module is used to upload the processed data set to the UAV flight control system, and the UAV flight control system supplements and fits the spatial point cloud data through spatial interpolation method;

[0062] A three-dimensional point cloud modeling module is used for the UAV flight control system to perform BIM modeling on the point cloud dataset through modeling technology, save it as an rvt file, and import it into a visualization platform for lightweight display;

[0063] The trajectory planning module is used by the UAV flight control system to plan the trajectory covering all detection points based on the constructed point cloud model, generate the trajectory planning file, and send it to the UAV;

[0064] The mission execution module is used for the UAV to receive the trajectory planning file and perform the inspection mission.

[0065] In a third aspect, an embodiment of the present application provides an electronic device, characterized by including:

[0066] processor;

[0067] a memory for storing processor-executable instructions;

[0068] In which, the processor is configured to implement the drone intelligent inspection method combining three-dimensional point cloud modeling and trajectory planning technology as described in the first aspect when executing the instructions.

[0069] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores a program, and the program instructs a device to execute the drone intelligent inspection method combining three-dimensional point cloud modeling and trajectory planning technology as described in the first aspect.

[0070] The present application provides a method and system for intelligent inspection of unmanned aerial vehicles (UAVs) that combines 3D point cloud modeling and trajectory planning technology, including data acquisition: scanning target detection points through a 3D laser scanning device on the UAV, performing point cloud data splicing, and saving the data as a point cloud dataset of a las file; data processing: performing Autodesk ReCap cutting on the point cloud dataset, performing format conversion, and saving the data as a point cloud dataset of an rcs file; uploading to a platform: uploading the processed data set to a UAV flight control system, which supplements and fits the spatial point cloud data through a spatial interpolation method; 3D point cloud modeling: the UAV flight control system performs BIM modeling on the point cloud dataset through modeling technology, saves the data as an rvt file, and imports the data into a visualization platform for lightweight display; trajectory planning: the UAV flight control system performs trajectory planning covering all detection points based on the constructed point cloud model, generates a trajectory planning file, and sends the file to the UAV; task execution: the UAV receives the trajectory planning file and performs the inspection task.

[0071] Beneficial effects:

[0072] (1) By combining 3D point cloud modeling with trajectory planning technology, the entire process is automated, forming a closed loop from data collection to trajectory generation, reducing manual intervention, and realizing the automation, refinement and intelligence of drone inspections, significantly improving inspection efficiency and safety.

[0073] (2) High-precision modeling and planning: processing point cloud data through pre-built point cloud data processing models to improve the quality of point cloud data. Combining BIM technology to achieve sub-centimeter-level modeling, the three-dimensional point cloud modeling has high accuracy, and the trajectory planning is based on real three-dimensional geometry to support refined inspection.

[0074] (3) By saving data in different file types, the transmission processing efficiency is improved, and route planning fully utilizes the spatial information of three-dimensional point clouds to support refined inspections.

[0075] (4) It is easy to promote and apply, and the operation is relatively simple, making it suitable for promotion and application in more application scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 A flowchart of a method for intelligent inspection by unmanned aerial vehicles (UAVs) that combines three-dimensional point cloud modeling with trajectory planning technology is provided in accordance with one embodiment of the present application.

[0077] Figure 2 This is a point cloud data processing model construction process provided in one embodiment of the present application.

[0078] Figure 3 A schematic diagram of the module of a drone intelligent inspection system that combines three-dimensional point cloud modeling and trajectory planning technology provided in one embodiment of the present application.

[0079] Figure 4 A schematic diagram of an electronic device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0080] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments.

[0081] It should be noted that, in the embodiments of the present application, "at least one" refers to one or more, and "more" refers to two or more. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art in the art to which this application relates. The terms used in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application.

[0082] It should be noted that, in the embodiments of the present application, words such as "first" and "second" are only used for the purpose of distinguishing descriptions, and cannot be understood as indicating or implying relative importance, nor can they be understood as indicating or implying an order. Features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or design schemes. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a concrete way.

[0083] Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0084] Example 1

[0085] This embodiment provides a method and system for intelligent inspection of unmanned aerial vehicles (UAVs) that combines three-dimensional point cloud modeling and trajectory planning technology, including: data collection, data processing, uploading to a platform, three-dimensional point cloud modeling, trajectory planning, and task execution modules. By combining three-dimensional point cloud modeling and trajectory planning technology, the entire process is automated, forming a closed loop from data collection to trajectory generation, reducing manual intervention, and realizing the automation, refinement, and intelligence of UAV inspections, significantly improving inspection efficiency and safety. High-precision modeling and planning, processing point cloud data through a pre-built point cloud data processing model to improve the quality of point cloud data, combining BIM technology to achieve sub-centimeter-level modeling, three-dimensional point cloud modeling with high accuracy, trajectory planning based on real three-dimensional geometry, and supporting refined inspections. By saving data in different file types, the transmission processing efficiency is improved, and route planning fully utilizes the spatial information of three-dimensional point clouds to support refined inspections.

[0086] Figure 1 A flowchart of a method for intelligent inspection by unmanned aerial vehicles (UAVs) that combines three-dimensional point cloud modeling with trajectory planning technology is provided in accordance with one embodiment of the present application.

[0087] like Figure 1 As shown, a UAV intelligent inspection method combining 3D point cloud modeling and trajectory planning technology includes:

[0088] S1: Data collection: Scan the target detection points using the 3D laser scanning device on the drone, perform point cloud data stitching, and save it as a point cloud dataset in the form of a LAS file.

[0089] High-precision point cloud data collection: Using lidar scanning technology, drones can acquire centimeter-level precision 3D point cloud data, ensuring clear visibility of details of inspection targets. As will be appreciated, in this embodiment, the drone's onboard 3D laser scanning equipment scans target inspection points, generating high-precision 3D point cloud data. This point cloud data typically covers the target area with centimeter-level accuracy, ensuring clear visibility of the tower structure, conductors, insulators, and key connections. After acquisition, the point cloud data is spliced and saved in a LAS file format for subsequent processing and analysis.

[0090] S1: Data Collection: Scan the target detection points using the 3D laser scanning device on the drone, perform point cloud data stitching, and save the point cloud data set as a LAS file, including:

[0091] Use the pre-built point cloud data processing model to filter and stitch the laser point cloud data.

[0092] Figure 2 This is a point cloud data processing model construction process provided in one embodiment of the present application. Figure 2 As shown, the point cloud data processing model is constructed by the following steps:

[0093] S1.1: Divide the target point cloud Q into a voxel grid and calculate the normal distribution of points within each voxel; optimize the probability scoring function to maximize the overlap between the projection of the source point cloud P and the distribution of Q;

[0094] S1.2: Use the corner detection algorithm to extract key points with obvious features by calculating the curvature change of the point cloud, and output the simplified point cloud P o and Q o ;

[0095] S1.3: P o For each point in Q o Find the points with the closest Euclidean distance as the corresponding point pair, solve the optimal R and T through SVD decomposition, and minimize the mean square error (MSE) of the corresponding point pairs. Stop when the average distance d is less than the threshold or reaches the maximum number of iterations.

[0096]

[0097] Among them, p i ∈P o ,q i ∈Q o ,i,n is the number of point cloud data sets, R is the matrix obtained by rotation change, and T is the matrix obtained by translation change;

[0098] S1.4: Randomly sample multiple candidate point pairs, use least squares to fit R and T, and select the model with the most point pairs whose statistical error is less than the threshold as the final registration result as the initial estimate.

[0099] S1.5: Adjust the point position by iterative nonlinear least squares method, and update the point position by iterative matrix J and matrix H, where

[0100]

[0101] H=J T J,

[0102]

[0103] p k+1 =p k +αΔp,

[0104] Among them, α represents the iterative step size of each update p, represents the mean value of the error;

[0105] S1.6: After the iteration, the point cloud data is stitched and saved as a point cloud dataset in a las file.

[0106] Specifically, three-dimensional laser scanning (LiDAR) is used to obtain high-precision point cloud data. The LAS format retains the geometry and attribute information (such as intensity and number of echoes) of the original scan, providing a reliable data basis for subsequent modeling.

[0107] S2: Data processing: The point cloud dataset is cut using Autodesk ReCap, format converted, and saved as a point cloud dataset in an rcs file.

[0108] It is understandable that in this embodiment, the collected point cloud data set is further processed, including cutting and format conversion using software such as Autodesk ReCap. The processed point cloud data is finally saved in rcs file format to facilitate subsequent modeling and analysis. Using tools such as ReCap to cut and convert the point cloud data, and then generating a three-dimensional model through BIM technology, improves data processing efficiency. Point cloud cutting (such as removing noise and segmenting areas of interest) is performed through Autodesk ReCap and converted to rcs format to optimize data volume and platform compatibility, adapting to subsequent BIM modeling requirements.

[0109] Specifically, the S2: data processing: performing Autodesk ReCap cutting on the point cloud dataset, performing format conversion, and saving the point cloud dataset as an rcs file, including:

[0110] Import point cloud data, open Autodesk ReCap software, and select "Open" from the "File" menu to import the LAS point cloud data file to be processed;

[0111] Cutting point cloud data: In ReCap, the user selects a specific area through the Limit Box to cut the point cloud data;

[0112] Clean and optimize data. After cutting is completed, use the cleaning tools provided by ReCap to remove useless data.

[0113] Convert to RCS format, and after completing the cutting and cleaning, export the RCS file and save it.

[0114] S3: Upload to the platform: upload the processed data set to the UAV flight control system, and the UAV flight control system supplements and fits the spatial point cloud data through the spatial interpolation method.

[0115] It is understood that in this embodiment, the processed point cloud data is uploaded to the management platform. The platform uses spatial interpolation to supplement and fit the point cloud data, thereby generating a more complete and accurate spatial model. Based on the three-dimensional point cloud model, combined with deep learning algorithms or improved A* search algorithms, safe and efficient inspection routes are automatically generated, reducing manual intervention. The flight control system uses spatial interpolation (such as kriging or inverse distance weighting) to fill in data gaps in scanning blind spots, enhance model integrity, and ensure that trajectory planning is complete.

[0116] Specifically, the S3: uploading to the platform: uploading the processed data set to the UAV flight control system, and the UAV flight control system supplements and fits the spatial point cloud data through the spatial interpolation method, including:

[0117] Preprocessing of raw point cloud data, including filtering, denoising, outlier detection, and data correction;

[0118] The improved inverse distance weighted method is used for spatial interpolation, and the formula is:

[0119]

[0120] Among them, let Z i is the attribute value at the reference point, u represents the index, d i Represents the Euclidean distance between the interpolation point p and the reference point. According to the characteristics of the interpolation function, when point p is infinitely close to the reference point, d i →0, the function value f1(p) will converge to Z i ; In particular, when d i = 0, f1(p) = Z i ;

[0121] The processed point cloud data is transmitted to the UAV flight control system via wireless signals. After receiving the data, the system uses spatial interpolation to supplement and fit the point cloud, and converts the discrete point cloud into a continuous surface through triangulation or rasterization technology. The ground point cloud data collected by the UAV is used to generate DEM and DSM, and the model is verified and evaluated for accuracy. It is combined with other sensor data for time alignment and spatial registration.

[0122] S4: 3D point cloud modeling: The UAV flight control system performs BIM modeling on the point cloud dataset through modeling technology, saves it as an rvt file, and imports it into a visualization platform for lightweight display.

[0123] It is understandable that in this embodiment, the management platform uses BIM modeling technology to perform three-dimensional modeling of the point cloud dataset and saves it in rvt file format. After these models are imported into the platform, they can be displayed in a lightweight manner, which facilitates the inspection personnel to intuitively understand the three-dimensional structure of the target area. The generated three-dimensional model is displayed in a lightweight manner on the platform, which is convenient for the inspection personnel to view and analyze in real time, and supports one-click inspection report generation. Based on the point cloud, a BIM model (rvt format) is generated and lightweight processing is performed through a visualization platform (such as Autodesk Forge or Unity3D), balancing model accuracy and real-time rendering efficiency.

[0124] Specifically, the S4: 3D point cloud modeling: the UAV flight control system performs BIM modeling on the point cloud dataset using modeling technology, saves it as an rvt file, and imports it into a visualization platform for lightweight display, including:

[0125] Use 3D modeling software to import the processed point cloud data and generate a preliminary 3D model, perform triangulation and texture mapping, and convert the discrete point cloud data into a continuous 3D surface model;

[0126] Based on the 3D model, the model is optimized through BIM modeling software, the attributes and parameters required by the Building Information Model (BIM) are added, the point cloud data is combined with the historical image data for analysis, deformation feature points are extracted, and a detailed BIM model is generated;

[0127] The final generated model is saved in .rvt format.

[0128] S5: Trajectory planning: The UAV flight control system performs trajectory planning covering all detection points based on the constructed point cloud model, generates a trajectory planning file, and sends it to the UAV.

[0129] Specifically, S5: trajectory planning: The UAV flight control system performs trajectory planning covering all detection points based on the constructed point cloud model, generates a trajectory planning file, and sends it to the UAV, including:

[0130] Through the 3D point cloud, the spatial information of key target points is extracted to generate a preliminary route;

[0131] According to the inspection objectives, parameters are designed for each point of interest, including shooting angle, distance, and height, and waypoints are planned in sequence according to the order of parameters;

[0132] Use Dijkstra algorithm to optimize path planning;

[0133] Add auxiliary points and set a reasonable safety distance between adjacent waypoints to prevent the drone from accidentally hitting the equipment during flight and ensure the safety of the route;

[0134] The generated route file should be stored in KML format for easy subsequent use.

[0135] It is understandable that in this embodiment, based on the generated three-dimensional point cloud model, route planning software is used for trajectory planning. During the planning process, factors such as environmental obstacles and safety distances are taken into account to generate one or more inspection routes. These routes can accurately guide the flight path of the drone within the target area, thereby achieving refined inspections. The trajectory planning algorithm (such as A, RRT or genetic algorithm) is combined with the spatial constraints of the point cloud model (obstacles, detection point distribution) to generate a collision-free, fully covered optimal path, thereby improving inspection efficiency.

[0136] Specifically, route planning can extract spatial information about inspection target locations from 3D point cloud data and, combined with route planning software, generate inspection routes on a 3D model. This approach improves the accuracy and efficiency of route planning, especially in complex environments. By combining point cloud data with the real-world model, model positions can be more accurately calibrated and routes generated that conform to the actual terrain. Furthermore, point cloud data can be used to analyze terrain features to optimize route paths.

[0137] Once route planning is complete, the system generates a trajectory planning file containing flight parameters (such as latitude and longitude, altitude, and speed). These files are typically saved in digital format and sent to the drone via a communication system. Upon receiving the planning file, the drone will perform flight simulation and execute the mission according to the preset trajectory parameters. For example, the 3D trajectory planning module combines the drone's flight characteristics, payload information, and environmental constraints to generate the optimal flight path.

[0138] The trajectory planning process is a core step in drone mission planning. Spatial information is extracted from a point cloud model and combined with route planning software to generate a trajectory file covering all inspection points. This file is then sent to the drone for mission execution. This process not only improves the accuracy and efficiency of route planning but also ensures safe flight in complex environments. The drone dynamically corrects its trajectory using real-time positioning (RTK-GNSS) and obstacle avoidance sensors (such as millimeter-wave radar) to ensure safe mission execution.

[0139] S6: Mission execution: The drone receives the trajectory planning file and performs the inspection mission.

[0140] After receiving the trajectory planning file, the drone will autonomously complete the inspection task according to the instructions in the file; during power inspections, the drone will fly according to the planned three-dimensional route to complete the specified aerial photography and data collection tasks; the drone's flight control software will adjust the flight attitude and trajectory in real time to adapt to environmental changes and maintain flight stability.

[0141] Example 2

[0142] like Figure 3 As shown, the present application provides a UAV intelligent inspection system that combines three-dimensional point cloud modeling and trajectory planning technology, which is applied to the UAV intelligent inspection method that combines three-dimensional point cloud modeling and trajectory planning technology as described in Example 1, including: It can be understood that in this embodiment, the data acquisition module 11, the data processing module 12, the data transmission module 13, the three-dimensional point cloud modeling module 14, the trajectory planning module 15, and the task execution module 16.

[0143] It can be understood that in this embodiment, the data acquisition module 11 is used to scan the target detection points through the three-dimensional laser scanning device on the drone, and to stitch the point cloud data and save it as a point cloud data set in a las file.

[0144] It can be understood that, in this embodiment, the data processing module 12 is used to perform Autodesk ReCap segmentation on the point cloud dataset, perform format conversion, and save the point cloud dataset as an rcs file.

[0145] It can be understood that, in this embodiment, the data transmission module 13 is used to upload the processed data set to the UAV flight control system, and the UAV flight control system supplements and fits the spatial point cloud data through the spatial interpolation method.

[0146] It can be understood that in this embodiment, the three-dimensional point cloud modeling module 14 is used for the UAV flight control system to perform BIM modeling on the point cloud data set through modeling technology, save it as an rvt file, and import it into the visualization platform for lightweight display.

[0147] It can be understood that, in this embodiment, the trajectory planning module 15 is used by the UAV flight control system to perform trajectory planning covering all detection points based on the constructed point cloud model, generate a trajectory planning file, and send it to the UAV.

[0148] It can be understood that, in this embodiment, the task execution module 16 is used for the UAV to receive the track planning file and perform the inspection task.

[0149] The present application provides a method and system for intelligent inspection of unmanned aerial vehicles (UAVs) that combines 3D point cloud modeling and trajectory planning technology, including data acquisition: scanning target detection points through a 3D laser scanning device on the UAV, performing point cloud data splicing, and saving the data as a point cloud dataset of a las file; data processing: performing Autodesk ReCap cutting on the point cloud dataset, performing format conversion, and saving the data as a point cloud dataset of an rcs file; uploading to a platform: uploading the processed data set to a UAV flight control system, which supplements and fits the spatial point cloud data through a spatial interpolation method; 3D point cloud modeling: the UAV flight control system performs BIM modeling on the point cloud dataset through modeling technology, saves the data as an rvt file, and imports the data into a visualization platform for lightweight display; trajectory planning: the UAV flight control system performs trajectory planning covering all detection points based on the constructed point cloud model, generates a trajectory planning file, and sends the file to the UAV; task execution: the UAV receives the trajectory planning file and performs the inspection task.

[0150] By combining 3D point cloud modeling with trajectory planning technology, the entire process is automated, forming a closed loop from data collection to trajectory generation, reducing manual intervention, and achieving automation, refinement, and intelligence for drone inspections, significantly improving inspection efficiency and safety. High-precision modeling and planning improve the quality of point cloud data by processing point cloud data through pre-built point cloud data processing models. Integrating BIM technology enables sub-centimeter modeling. 3D point cloud modeling is highly accurate, and trajectory planning is based on real 3D geometry, supporting refined inspections. Saving data in different file types improves transmission processing efficiency, and route planning fully utilizes the spatial information of 3D point clouds to support refined inspections. Easy to promote and apply, and relatively simple to operate, it is suitable for promotion and application in a wider range of application scenarios.

[0151] Figure 4 This is an electronic device provided by an embodiment of the present application. Figure 4 As shown, the electronic device includes at least the following parts: a processor 101 and a memory 100 , a communication interface 103 , and a bus 102 .

[0152] In the embodiment of the present application, the memory 100 is used to store instructions executable by the processor 101. The processor 101 is configured to execute the instructions to implement the following Figure 3 The device module shown is for UAV intelligent inspection that combines 3D point cloud modeling and trajectory planning technology.

[0153] In an embodiment of the present application, a computer-readable storage medium includes instructions, and the instructions instruct a device to execute the method of the first aspect. For example, the instructions instruct the device to execute Figure 1 The method is shown in the process steps.

[0154] The program running in the electronic device involved in one embodiment of the present application can be a program that controls a central processing unit (CPU) and the like to realize the functions of the above-mentioned embodiment involved in one embodiment of the present invention (a program that enables a computer to function). Then, the information processed by these devices is temporarily stored in a random access memory (RAM) during its processing, and then stored in various ROMs such as read-only memory (Flash ROM) and hard disk drive (HDD), and is read, modified, and written by the CPU as needed.

[0155] It should be noted that a portion of the electronic device of the above embodiment may also be implemented by a computer. In this case, a program for implementing the control function may be recorded on a computer-readable recording medium, and the program recorded on the recording medium may be read into a computer and executed.

[0156] It should be noted that the "computer" mentioned here refers to a computer built into an electronic device, employing hardware including an operating system (OS) and peripheral devices. Furthermore, "computer-readable recording medium" refers to removable media such as floppy disks, magneto-optical disks, ROMs, and CD-ROMs, as well as storage devices such as hard disks built into computers.

[0157] Furthermore, "computer-readable recording media" may include: media that dynamically store programs for a short period of time, such as communication lines when transmitting programs via networks such as the Internet or communication lines such as telephone lines; and media that store programs for a fixed period of time, such as volatile memory within computers acting as servers or clients in this context. Furthermore, the aforementioned program may be a program for implementing a portion of the aforementioned functions, or a program that can achieve the aforementioned functions by combining with a program already stored in a computer.

[0158] Furthermore, the electronic device in the above-described embodiments can also be implemented as a collection (device group) consisting of multiple devices. Each device constituting the device group may have a portion or all of the functions or functional blocks of the electronic device in the above-described embodiments. A device group only needs to have all the functions or functional blocks of the electronic device.

[0159] Those skilled in the art should recognize that the above embodiments are merely intended to illustrate the present application and are not intended to limit the present application. As long as they are within the spirit of the present application, appropriate changes and modifications to the above embodiments are within the scope of protection claimed in the present application.

Claims

1. A UAV intelligent inspection method combining 3D point cloud modeling and trajectory planning technology, characterized in that: The method comprises: S1: Data collection: Scan the target detection points using the 3D laser scanning device on the drone, perform point cloud data stitching, and save it as a point cloud dataset in a LAS file; S2: Data processing: Perform Autodesk ReCap cutting on the point cloud dataset, convert the format, and save it as a point cloud dataset of rcs file; S3: Upload to the platform: upload the processed data set to the UAV flight control system, and the UAV flight control system supplements and fits the spatial point cloud data through spatial interpolation method; S4: 3D point cloud modeling: The UAV flight control system performs BIM modeling on the point cloud dataset through modeling technology, saves it as an rvt file, and imports it into a visualization platform for lightweight display; S5: Trajectory planning: The UAV flight control system plans a trajectory covering all detection points based on the constructed point cloud model, generates a trajectory planning file, and sends it to the UAV; S6: Mission execution: The drone receives the trajectory planning file and performs the inspection mission.

2. The method for intelligent inspection of unmanned aerial vehicles (UAVs) combining 3D point cloud modeling and trajectory planning technology according to claim 1, characterized in that: S1: Data Collection: Scan the target detection points using the 3D laser scanning device on the drone, perform point cloud data stitching, and save the point cloud data set as a LAS file, including: Use pre-built point cloud data processing models to filter and stitch laser point cloud data; The point cloud data processing model is constructed by the following steps: S1.1: Divide the target point cloud Q into a voxel grid and calculate the normal distribution of points within each voxel; optimize the probability scoring function to maximize the overlap between the projection of the source point cloud P and the distribution of Q; S1.2: Use the corner detection algorithm to extract key points with obvious features by calculating the curvature change of the point cloud, and output the simplified point cloud P o and Q o ; S1.3: P o For each point in Q o Find the points with the closest Euclidean distance as the corresponding point pair, solve the optimal R and T through SVD decomposition, and minimize the mean square error (MSE) of the corresponding point pairs. Stop when the average distance d is less than the threshold or reaches the maximum number of iterations. Among them, p i ∈P o ,q i ∈Q o ,i,n is the number of point cloud data sets, R is the matrix obtained by rotation change, and T is the matrix obtained by translation change; S1.4: Randomly sample multiple candidate point pairs, use least squares to fit R and T, and select the model with the most point pairs whose statistical error is less than the threshold as the final registration result as the initial estimate. S1.5: Adjust the point position by iterative nonlinear least squares method, and update the point position by iterative matrix J and matrix H, where H=J T J, p k+1 =p k +αΔp, Among them, α represents the iterative step size of each update p, represents the mean value of the error; S1.6: After the iteration, the point cloud data is stitched and saved as a point cloud dataset in a las file.

3. The UAV intelligent inspection method combining 3D point cloud modeling and trajectory planning technology according to claim 2 is characterized in that: S2: Data processing: Performing Autodesk ReCap cutting on the point cloud dataset and performing format conversion to save the point cloud dataset as an rcs file, including: Import point cloud data. Open Autodesk ReCap software and select "Open" from the "File" menu to import the LAS point cloud data file to be processed. Cutting point cloud data: In ReCap, the user selects a specific area through the Limit Box to cut the point cloud data; Clean and optimize data. After cutting is completed, use the cleaning tools provided by Re Cap to remove useless data. Convert to RCS format, and after completing the cutting and cleaning, export the RCS file and save it.

4. The method for intelligent inspection of unmanned aerial vehicles (UAVs) combining 3D point cloud modeling and trajectory planning technology according to claim 2, characterized in that: S3: Upload to the platform: Upload the processed data set to the UAV flight control system, and the UAV flight control system supplements and fits the spatial point cloud data through the spatial interpolation method, including: Preprocessing of raw point cloud data, including filtering, denoising, outlier detection, and data correction; The improved inverse distance weighted method is used for spatial interpolation, and the formula is: Among them, let Z i is the attribute value at the reference point, u represents the index, d i Represents the Euclidean distance between the interpolation point p and the reference point. According to the characteristics of the interpolation function, when point p is infinitely close to the reference point, d i →0, the function value f1(p) will converge to Z i ; In particular, when d i = 0, f1(p) = Z i ; The processed point cloud data is transmitted to the UAV flight control system via wireless signals. After receiving the data, the system uses spatial interpolation to supplement and fit the point cloud, and converts the discrete point cloud into a continuous surface through triangulation or rasterization technology. The ground point cloud data collected by the UAV is used to generate DEM and DSM, and the model is verified and evaluated for accuracy. It is combined with other sensor data for time alignment and spatial registration.

5. The method for intelligent inspection of unmanned aerial vehicles combining three-dimensional point cloud modeling and trajectory planning technology according to claim 2, characterized in that: S4: 3D point cloud modeling: The UAV flight control system performs BIM modeling on the point cloud dataset using modeling technology, saves it as an rv t file, and imports it into a visualization platform for lightweight display, including: Use 3D modeling software to import the processed point cloud data and generate a preliminary 3D model, perform triangulation and texture mapping, and convert the discrete point cloud data into a continuous 3D surface model; Based on the 3D model, the model is optimized through BIM modeling software, the attributes and parameters required by the Building Information Model (BIM) are added, the point cloud data is combined with the historical image data for analysis, deformation feature points are extracted, and a detailed BIM model is generated; The final generated model is saved in .rvt format.

6. The UAV intelligent inspection method combining 3D point cloud modeling and trajectory planning technology according to claim 3 is characterized in that: S5: Track planning: The UAV flight control system performs track planning covering all detection points based on the constructed point cloud model, generates a track planning file, and sends it to the UAV, including: Through the 3D point cloud, the spatial information of key target points is extracted to generate a preliminary route; According to the inspection objectives, parameters are designed for each point of interest, including shooting angle, distance, and height, and waypoints are planned in sequence according to the order of parameters; Use Dijkstra algorithm to optimize path planning; Add auxiliary points and set a reasonable safety distance between adjacent waypoints to prevent the drone from accidentally hitting the equipment during flight and ensure the safety of the route; The generated route file should be stored in KML format for easy subsequent use.

7. The UAV intelligent inspection method combining 3D point cloud modeling and trajectory planning technology according to claim 5 is characterized in that: The method of extracting spatial information of key target points through the three-dimensional point cloud and generating a preliminary route includes: Extract the spatial information of the target point through 3D point cloud data, and generate a preliminary route using route planning software; Before said generated route file should be stored in KML format include: According to inspection requirements, set the shooting angle, pitch angle, shooting distance and hovering time parameters for each point of interest; Arrive at waypoints in the predetermined order and perform photo-taking tasks to ensure that all inspection targets are covered; After the route planning is completed, the route is rehearsed in a simulated environment to check for safety hazards or missed areas. Waypoints and photography parameters are adjusted through real-time simulation to ensure the safety and integrity of the route. Manual review and fine-tuning, based on actual on-site conditions, fine-tune the waypoint positions and parameters to ensure that the route meets actual needs.

8. An intelligent inspection system for unmanned aerial vehicles (UAVs) that combines 3D point cloud modeling and trajectory planning technology, applied to the intelligent inspection method for unmanned aerial vehicles (UAVs) that combines 3D point cloud modeling and trajectory planning technology as described in any one of claims 1 to 7, characterized in that: include: The data acquisition module is used to scan the target detection points through the 3D laser scanning device on the UAV, stitch the point cloud data, and save it as a point cloud dataset in the form of a LAS file; A data processing module is used to perform Autodesk ReCap cutting on the point cloud dataset, convert the format of the point cloud dataset, and save it as an rcs file; A data transmission module is used to upload the processed data set to the UAV flight control system, and the UAV flight control system supplements and fits the spatial point cloud data through spatial interpolation method; A three-dimensional point cloud modeling module is used for the UAV flight control system to perform BIM modeling on the point cloud dataset through modeling technology, save it as an rvt file, and import it into a visualization platform for lightweight display; The trajectory planning module is used by the UAV flight control system to plan the trajectory covering all detection points based on the constructed point cloud model, generate the trajectory planning file, and send it to the UAV; The mission execution module is used for the UAV to receive the trajectory planning file and perform the inspection mission.

9. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; Wherein, the processor is configured to implement the UAV intelligent inspection method combining three-dimensional point cloud modeling and trajectory planning technology as described in any one of claims 1 to 7 when executing the instructions.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a program, and the program instructs the device to execute the drone intelligent inspection method combining three-dimensional point cloud modeling and trajectory planning technology as described in any one of claims 1 to 7.

Images