Bridge disease identification method based on low-altitude unmanned aerial vehicle

Abstract

The application relates to the technical field of bridge health monitoring, in particular to a bridge disease identification method based on a low-altitude unmanned aerial vehicle, which comprises the following steps: acquiring a bridge model and extracting component three-dimensional space and attribute features; dividing global and fine detection areas and configuring constraint parameters; generating an initial inspection route, wherein the fine area comprises a close-to-edge path and multi-angle nodes; controlling the unmanned aerial vehicle to fly and acquire real-time environmental data; fusing and comparing the data with the model, dynamically adjusting the route when a triggering condition is met; and collecting images according to the adjusted route to identify diseases. In the application, a differentiated and self-adaptive route planning and execution mechanism based on a building information model is constructed, technical problems that the route mode of the prior art is single, complex components cannot be covered, and the environment cannot be adapted to dynamic changes are solved, and the application has the beneficial effect of being capable of realizing non-blind area and high-precision coverage of all components of a bridge, especially complex nodes, and significantly improving the safety of inspection flight.

Description

Technical Field

[0001] This invention relates to the field of bridge health monitoring technology, and in particular to a method for identifying bridge defects based on low-altitude unmanned aerial vehicles (UAVs). Background Technology

[0002] As a crucial component of transportation infrastructure, the structural safety and daily maintenance of bridges are of paramount importance. Utilizing low-altitude drones equipped with high-definition imaging equipment to identify defects in all bridge components has become a primary technical means for obtaining inspection data in the field of bridge health monitoring, thanks to its flexibility and close-range operation advantages. It plays an irreplaceable role, especially in the detailed inspection of multiple complex components such as supports, cables, and towers of large cable-stayed bridges and suspension bridges.

[0003] Currently, in existing technologies, drones typically fly automatically along preset flight paths and trigger image capture. Specifically, this involves: first, manually pre-setting a set of latitude and longitude waypoints based on the bridge's design drawings or general outline, or generating a flight path using a single surround photography or orthophoto projection mode; then, guided by satellite positioning signals, the drone maintains a constant flight altitude and speed to complete the image acquisition task.

[0004] However, the aforementioned existing technologies have the following drawbacks in practical applications: Firstly, the lack of targeted area segmentation logic and the use of a single, coarse-grained flight path pattern make them unsuitable for the spatial characteristics of complex structures such as bridge underpasses, supports, and cable nodes. This hinders the UAV from achieving accurate close-range, multi-angle imaging, turning these high-risk areas into blind spots. Secondly, the lack of differentiated flight path generation and the globally uniform flight parameters result in blurred imaging details of complex components, failing to meet the accuracy requirements for defect identification at the ≤0.1mm level. Furthermore, redundant data is generated in non-core areas, reducing inspection efficiency. Thirdly, existing methods lack an adaptive adjustment mechanism driven by a high-precision model. When there are construction deviations in the actual position of bridge components or temporary obstacles in the flight path, the preset flight path cannot be dynamically corrected in real time, increasing the risk of missed detections and compromising flight safety in complex low-altitude environments.

[0005] Therefore, this application aims to solve the problems of the existing bridge inspection route planning mode being too simple and insufficient coverage of complex components, and also to solve the problem that the route cannot be dynamically adjusted in real time according to the actual position and orientation characteristics of the components and environmental obstacles. Summary of the Invention

[0006] To address the technical problems existing in the prior art, this invention provides a method for identifying bridge defects based on low-altitude unmanned aerial vehicles (UAVs), comprising the following steps: To achieve the above objectives, the present invention adopts the following technical solution: a bridge defect identification method based on low-altitude unmanned aerial vehicles, comprising the following steps: S1: Obtain the architectural information model of the target bridge, and extract the three-dimensional spatial features and attribute features of the bridge components of the target bridge; S2: Based on the three-dimensional spatial features and attribute features, the target bridge is divided into a global detection area and a fine detection area, and corresponding inspection constraint parameters are configured for different detection areas; S3: Generate corresponding initial inspection routes based on the three-dimensional spatial features of the global detection area and the fine detection area; wherein, for the fine detection area, a fine inspection route including edge-following flight path and multi-angle shooting nodes is generated. S4: Control the low-altitude UAV to fly along the initial inspection route and acquire real-time environmental perception data during the flight; S5: Spatially fuse and compare the real-time environmental perception data with the building information model. When the pre-set trigger adjustment conditions are met, the initial inspection route is dynamically adjusted under the premise of prioritizing the inspection constraint parameters. S6: Fly along the adjusted inspection route and collect bridge image data for bridge defect identification.

[0007] As a further aspect of the present invention, the extraction of the three-dimensional spatial features and attribute features of the bridge components included in the target bridge includes: The building information model is simplified geometrically, and non-core detailed components are removed. Extract the three-dimensional contour coordinates, spatial distribution relationship, and disease susceptibility level of the retained bridge components, and establish the mapping relationship between component type and the three-dimensional spatial features and attribute features.

[0008] As a further aspect of the present invention, based on the three-dimensional spatial features and attribute features, the target bridge is divided into a global detection region and a fine detection region, and corresponding inspection constraint parameters are configured for different detection regions, including: Identify exposed and regularly shaped bridge components in the building information model, and divide the spatial range of their location into the global detection area; Identify components in the building information model that are hidden, structurally irregular, and have a susceptibility level to defects higher than a preset threshold, and divide the spatial range of these components into the fine detection area. Based on the type and size of different components within the fine detection area, the inspection constraint parameters are generated. The inspection constraint parameters include at least: target shooting distance range, shooting angle coverage range, and flight speed limit.

[0009] As a further aspect of the present invention, the step of generating corresponding initial inspection routes based on the three-dimensional spatial features of the global detection area and the fine detection area includes: For the global detection area, a global inspection route is generated based on the outer contour of the main structure of the target bridge, which is used to fly around the target bridge and satisfy the preset heading and lateral overlap rate. For the fine inspection area, based on the three-dimensional contour coordinates of the target component, a path trajectory parallel to the outer surface of the component is extracted to generate an edge-following flight path whose distance from the component surface is within the target shooting distance range; according to the three-dimensional structural morphology characteristics of the target bridge component in the fine inspection area, multiple key shooting angle nodes that meet the shooting angle coverage range are set on the edge-following flight path to generate the fine inspection route.

[0010] As a further aspect of the present invention, the step of spatially fusing and comparing the real-time environmental perception data with the building information model, and dynamically adjusting the initial inspection route when a preset trigger adjustment condition is met, includes: If the real-time environmental perception data indicates that there is an obstacle ahead of the flight path and the distance to the initial inspection route is less than the preset safe distance, then the obstacle avoidance adjustment condition is triggered; a local detour path is planned based on the pathfinding algorithm, and the inspection constraint parameters are used as the forced constraint nodes of the pathfinding algorithm, so that the local detour path can avoid obstacles while satisfying the set edge distance and shooting angle. If the real-time environmental perception data indicates that the displacement deviation between the actual spatial position of the bridge component and the three-dimensional spatial coordinates in the building information model is greater than or equal to a preset deviation threshold, then the registration adjustment condition is triggered; the real-time point cloud data is spatially registered with the building information model using a point cloud registration algorithm, and the waypoint coordinates in the initial inspection route are synchronously corrected based on the registration offset.

[0011] As a further aspect of the present invention, the step of flying along the adjusted inspection route and collecting bridge image data includes: When the low-altitude UAV is in the global detection area, it triggers shooting at equal intervals according to the first preset flight speed; When the low-altitude UAV is in the fine inspection area, it flies at a second preset flight speed, which is lower than the first preset flight speed. When it reaches the multi-angle shooting node set in the fine inspection route, it controls the camera attitude and triggers shooting at the second preset shooting frequency.

[0012] As a further aspect of the present invention, the acquisition of real-time environmental perception data includes: Simultaneously acquire depth point cloud data from airborne lidar and image pixel data from visual sensors; The depth point cloud data and the image pixel data are time-stamp aligned and spatial coordinate calibrated. The calibrated data is fused to construct a real-time 3D point cloud map, and the 3D boundary contours of potential obstacles are extracted based on the real-time 3D point cloud map.

[0013] As a further aspect of the present invention, the dynamic adjustment of the initial inspection route also includes an anomaly handling step: When the pathfinding algorithm determines that a local detour path cannot be generated under the constraints of the target shooting distance range and the shooting angle coverage range, the low-altitude UAV is controlled to hover and the current route interruption point is recorded. A reverse safe evacuation trajectory is generated based on the location of the flight route interruption, and the bridge components corresponding to the location of the flight route interruption are marked as requiring manual re-inspection in the building information model.

[0014] As a further aspect of the present invention, the bridge defect identification includes: Extract two-dimensional features of bridge defects from the collected bridge image data; Based on the spatial pose data of the low-altitude UAV and the intrinsic parameter matrix of the camera when collecting the bridge image data, the two-dimensional features of the defects are mapped back to the three-dimensional spatial coordinate system of the building information model. The building information model automatically generates label information including the type of disease, the three-dimensional coordinates of the disease, and the size of the disease.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, before UAV inspection, the architectural information model of the target bridge is acquired and the three-dimensional spatial and attribute features of the components are extracted. Based on these features, the bridge is divided into two types of inspection areas: global and fine, and differentiated inspection constraint parameters are configured. Then, initial inspection routes matching the spatial features of different areas are generated. In particular, routes with edge-following flight paths and multi-angle shooting nodes are generated for fine inspection areas. Real-time environmental data is integrated during flight to dynamically adjust the routes. Images are collected according to the adjusted routes. This enables UAV route planning to change from extensive and indiscriminate to refined and adaptive, thereby solving the problem in the background technology where the single route mode leads to insufficient coverage of multiple complex components under the bridge and supports and blurred imaging details. This achieves high-precision detection of bridge defects in high-incidence areas without blind spots.

[0016] Furthermore, by establishing a dynamic adjustment mechanism based on spatial fusion and comparison of real-time environmental perception data and building information models, the flight path can be corrected in real time when temporary obstacles or construction deviations in the actual positions of bridge components occur during flight. This design transforms the inspection route from a rigid preset trajectory into one with the ability to autonomously avoid obstacles and achieve precise registration in complex environments. This significantly improves flight safety for low-altitude operations and ensures that the flight path flying close to the edge always precisely matches the actual position of the component, solving the problems of preset routes being unable to be dynamically corrected, resulting in missed inspections and safety hazards in the prior art.

[0017] Furthermore, by generating a flight path along the edge based on the three-dimensional contour coordinates of the target component within the fine detection area, and setting multiple key shooting angle nodes according to its structural morphology characteristics, it is possible to ensure that the UAV can take close-up shots of multiple components with high incidence of defects but complex structures, such as supports and cable nodes, from multiple key perspectives at extremely close distances, and acquire high-definition images that meet the requirements for defect identification accuracy of ≤0.1mm level. This solves the problem in the background technology where the imaging of complex components is blurred due to globally uniform flight parameters, making it difficult to meet the requirements for detailed detection.

[0018] Furthermore, after acquiring image data, based on the UAV's spatial pose data and camera intrinsic parameters, the two-dimensional defect features in the images are mapped back to the three-dimensional spatial coordinate system of the building information model, and defect label information containing three-dimensional coordinates is automatically generated. This enables the accurate location and quantification of discovered defects in the digital model, providing intuitive and accurate three-dimensional data support for subsequent bridge structural safety assessments and the formulation of maintenance and reinforcement plans, thereby improving the digital level of bridge life-cycle health management. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in 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.

[0020] Figure 1 This is a schematic diagram of the steps of the present invention. Detailed Implementation

[0021] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0022] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0023] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0024] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0025] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0026] In the technical solution of this application, the Building Information Model (BIM) is a digital model that includes the type, three-dimensional spatial coordinates, dimensional parameters, structural form, and disease susceptibility level of all bridge components, supporting common data formats such as IFC and Revit. The three-dimensional spatial features refer to multiple geometrically related features of the bridge components, including their three-dimensional contour coordinates, spatial distribution relationships, and structural form. The attribute features refer to multiple non-spatial geometrically related features of the bridge components, including their type, disease susceptibility level, and dimensional parameters. The inspection constraint parameters are restrictive parameters set for UAV flight and photography to adapt to the disease identification needs of different inspection areas. The edge-following flight path refers to the UAV flight path that maintains a preset distance from the outer surface of the bridge components and conforms to the surface contour of the components. The trigger adjustment conditions refer to the judgment conditions that require dynamic adjustment of the initial inspection route during UAV flight, including obstacle avoidance adjustment conditions and registration adjustment conditions.

[0027] Please see Figure 1 This invention provides a method for identifying bridge defects based on low-altitude unmanned aerial vehicles (UAVs), comprising the following steps: S1: Obtain the architectural information model of the target bridge and extract the three-dimensional spatial features and attribute features of the bridge components of the target bridge; Obtain the IFC format building information model of the target bridge. This model contains complete information on all bridge components, including the bridge deck, piers, bearings, cables, U-ribs, and towers. First, the building information model is simplified geometrically by removing decorative components, attached green structures, redundant geometric surfaces, and non-core details. This reduces the redundancy in subsequent data processing and improves the efficiency of area division and route generation. After model simplification, the three-dimensional contour coordinates, spatial distribution relationships, and disease susceptibility levels of the retained bridge components are extracted. At the same time, a mapping relationship between component type and three-dimensional spatial features and attribute features is established, forming an associated database of component type, spatial coordinates, structural morphology, and disease susceptibility. The purpose of this step is to provide accurate and effective data support for subsequent inspection area division, inspection constraint parameter configuration, and inspection route generation, ensuring that the implementation of subsequent steps matches the actual structural characteristics of the bridge and the needs of disease identification. S2: Based on three-dimensional spatial features and attribute features, the target bridge is divided into a global detection area and a fine detection area, and corresponding inspection constraint parameters are configured for different detection areas; Using the associated database obtained in step S1 as a basis, the detection area is divided by combining the three-dimensional spatial features and attribute features of bridge components. Specifically, the exposed and regularly shaped bridge components in the building information model, such as the bridge deck and the outer side of the pier, are identified. These components have a low susceptibility to defects and high accessibility for manual inspection. Their spatial range is divided into the global detection area. The core target of the detection in this area is the integrity of the coverage, which does not require close-range shooting. Identify components in the building information model that are hidden, structurally irregular, and have a disease susceptibility level higher than a preset threshold, such as supports, cables and nodes, U-ribs, bridge bottoms, and the inner side of pylons. These components are high-incidence areas for bridge diseases and are difficult to access manually. Divide the space in which they are located into a fine detection area. This area is the core area for disease identification and requires close-range, multi-angle, and precise photography. After the area is divided, inspection constraint parameters are generated based on the type and size of different components within the fine inspection area. The inspection constraint parameters include at least the target shooting distance range, the shooting angle coverage range, and the upper limit of flight speed. In this embodiment, the inspection constraint parameters configured for the fine inspection area are a shooting distance of 0.5-2m, a shooting angle coverage of 30-60 degrees downward angle, and an upper limit of flight speed of 1.5m / s. The global inspection area does not need to be configured with dedicated inspection constraint parameters, and only needs to meet the normal flight and shooting requirements.

[0028] The purpose of this step is to achieve accurate and differentiated division of the detection area, configure corresponding constraint parameters for the structural characteristics and disease identification needs of different areas, provide a basis for the generation of subsequent differentiated inspection routes, and solve the problems of non-differentiated route planning and insufficient coverage of complex structures in traditional methods. S3: Generate corresponding initial inspection routes based on the three-dimensional spatial features of the global detection area and the fine detection area; among them, generate fine inspection routes for the fine detection area that include edge-following flight paths and multi-angle shooting nodes. Based on the three-dimensional spatial features of the global detection area and the fine detection area, a differentiated strategy is adopted to generate the corresponding initial inspection route. The specific implementation method is as follows: For the global detection area, a global inspection route is generated based on the outer contour of the target bridge's main structure, which is used to fly around the area and meet the preset heading and lateral overlap rates. In this embodiment, the global inspection route is a low-altitude circling + oblique photography composite route along the length of the bridge. The flight altitude is set to 5-30m, the flight speed is 2-3m / s, the heading overlap rate is ≥70%, and the lateral overlap rate is ≥80%. The purpose of this design is to ensure that the global detection area is fully covered while avoiding collisions between the UAV and the main bridge structure during low-altitude flight, thus balancing coverage integrity and flight safety. For the fine detection area, based on the three-dimensional contour coordinates of the target component, the path trajectory parallel to the outer surface of the component is extracted, and an edge-following flight path is generated with the distance from the component surface within the target shooting distance range; Based on the three-dimensional structural features of the target bridge components within the fine inspection area, multiple key shooting angle nodes that meet the shooting angle coverage range are set on the edge-following flight path to generate a fine inspection route.

[0029] In this embodiment, for the support component, the tangent trajectory of the outer surface of the support is extracted as the edge-following flight path. The distance between the path and the support surface is maintained at 0.8-1.5m, and four key shooting angle nodes are set on the path: horizontal angle, downward angle of 20-30 degrees, upward angle of 40-60 degrees, and lateral angle of 45 degrees. For cable components, a double-track edge-flying path parallel to the cable is planned, with the double tracks located 0.5-1m on each side of the cable. Three key shooting angle nodes, namely axial angle, radial angle, and nodal oblique angle, are set every 5-10m along the length of the cable. The purpose of this step is to generate suitable initial inspection routes for different inspection areas. For the fine inspection areas of complex structures, the edge-flying path and multi-angle shooting nodes are designed to ensure that the UAV flies close to the surface of the component, realizes the shooting of complex structures without blind spots, and solves the problem of high missed detection rate of complex structure defects in traditional methods. S4: Control the low-altitude UAV to fly along the initial inspection route and acquire real-time environmental perception data during the flight; The low-altitude UAV is controlled to start flying according to the initial inspection route generated in step S3. During the flight, real-time environmental perception data is collected synchronously through the airborne lidar and binocular vision sensor. Specifically, the depth point cloud data of the airborne lidar and the image pixel data of the vision sensor are collected synchronously. The depth point cloud data and image pixel data are time-stamped and spatially calibrated to eliminate the time difference and spatial position deviation of the acquisition of different sensors. The calibrated data are then feature-fused to construct a real-time three-dimensional point cloud map, and the three-dimensional boundary contours of potential obstacles are extracted based on the real-time three-dimensional point cloud map. The purpose of this step is to acquire spatial information of the UAV's flight environment in real time, accurately identify potential obstacles on the flight path, and provide a data basis for the subsequent comparison of the actual location of bridge components with the building information model, so as to ensure the timeliness and accuracy of dynamic adjustments to the flight path. S5: Spatially fuse and compare real-time environmental perception data with building information model. When the pre-set trigger adjustment conditions are met, the initial inspection route is dynamically adjusted under the premise of prioritizing the inspection constraint parameters. The real-time environmental perception data obtained in step S4 is spatially fused with the building information model of the target bridge to establish a three-dimensional spatial relationship between the flight path, obstacles, and bridge components. The flight status of the UAV and the actual position of the bridge components are determined in real time. When the pre-set trigger adjustment conditions are met, the initial inspection route is dynamically adjusted under the premise of prioritizing the satisfaction of inspection constraint parameters. The specific determination and adjustment methods are as follows: If real-time environmental perception data indicates that there is an obstacle ahead of the flight path and the distance from the initial inspection route is less than the preset safe distance (in this embodiment, the safe distance is set to 1m), then the obstacle avoidance adjustment condition is triggered. Based on the improved A* pathfinding algorithm, a local detour path is planned, and the inspection constraint parameters are used as the mandatory constraint nodes of the pathfinding algorithm. This ensures that the local detour path meets the set edge distance and shooting angle while avoiding obstacles. After the detour is completed, the UAV automatically returns to the initial inspection route. The purpose of this design is to ensure that the UAV can still meet the shooting and flight constraints of the fine inspection area while avoiding obstacles, and to avoid the failure of shooting complex structures due to obstacle avoidance.

[0030] If the real-time environmental perception data indicates that the displacement deviation between the actual spatial position of the bridge component and the three-dimensional spatial coordinates in the building information model is greater than or equal to a preset deviation threshold (in this embodiment, the deviation threshold is set to 3cm), then the registration adjustment condition is triggered. The ICP point cloud registration algorithm is used to spatially register real-time point cloud data with the building information model, and the waypoint coordinates in the initial inspection route are synchronously corrected based on the registration offset to ensure that the relative position between the edge-flying path and the component surface remains unchanged. The purpose of this design is to correct the discrepancy between the building information model and the actual structure of the bridge, ensuring that the inspection route always accurately matches the actual position of the components and guaranteeing the accuracy of the images. When the pathfinding algorithm determines that a local detour path cannot be generated under the constraints of the target shooting distance range and the shooting angle coverage range, the anomaly handling step is initiated. The low-altitude UAV is controlled to hover and record the current route interruption location. Then, a reverse safe evacuation trajectory is generated based on the route interruption location to ensure the safe evacuation of the UAV. At the same time, the bridge component corresponding to the route interruption location is marked as requiring manual re-inspection in the building information model. The purpose of this step is to deal with extreme obstacle avoidance scenarios, ensure the flight safety of the drone when automatic obstacle avoidance detection is not possible, and mark undetected components to avoid blind spots in defect identification. The overall purpose of this step is to enable dynamic adjustment of the inspection route, allowing the UAV to adapt to complex actual flight environments, while taking into account flight safety, shooting accuracy and detection coverage, and solving the problem that the traditional method has a fixed route and cannot adapt to changes in the actual environment. S6: Fly along the adjusted inspection route and collect bridge image data for bridge defect identification; The low-altitude drone is controlled to fly along the inspection route adjusted in step S5. During the flight, a differentiated shooting strategy is adopted to collect bridge image data according to the different detection areas. Specifically, when the low-altitude drone is in the global detection area, it is triggered to shoot at equal intervals according to the first preset flight speed of 2-3 m / s. In this embodiment, the shooting interval is set to 1 second. When the low-altitude UAV is in the fine inspection area, it flies at a second preset flight speed of 1-1.5 m / s, which is lower than the first preset flight speed. When it reaches the multi-angle shooting node set in the fine inspection route, it controls the camera attitude to match the shooting angle and triggers shooting according to the second preset shooting frequency. In this embodiment, the shooting interval is set to ≤0.5s. The purpose of this imaging strategy is to balance detection efficiency and imaging accuracy. The global detection area ensures coverage efficiency, while the fine detection area ensures clear imaging of disease details through low-speed flight and high-frequency imaging.

[0031] After the data collection is completed, the bridge image data is processed to identify bridge defects. Specifically, the two-dimensional features of defects in the collected bridge image data are extracted, such as the shape, size, and location of cracks, corrosion, and damage. Then, based on the spatial pose data of the low-altitude UAV and the camera's intrinsic parameter matrix when collecting bridge image data, the two-dimensional features of the defects are mapped back to the three-dimensional spatial coordinate system of the building information model. Finally, the building information model automatically generates label information including disease type, disease 3D coordinates, and disease size. The purpose of this processing method is to transform the 2D image disease features into 3D spatial disease information, so as to achieve accurate positioning and quantitative labeling of the disease.

[0032] This embodiment represents the optimal implementation of a bridge defect identification method based on low-altitude UAVs. It is applicable to the identification of defects in all components of various beam bridges, arch bridges, cable-stayed bridges, and suspension bridges, including highways and railways. It is particularly suitable for the precise detection of multiple complex structures such as supports and cables. In this embodiment, a multi-rotor low-altitude UAV is selected as the execution carrier. The UAV is equipped with an RTK-GNSS positioning module, a lidar, a binocular vision sensor, and a 4K high-definition industrial camera. The positioning module has a dynamic positioning accuracy of ±2cm, the lidar can detect obstacles with a diameter ≥0.5cm within a range of 0.1-50m, and the high-definition industrial camera has a pixel count of no less than 20 million, which can meet the shooting accuracy requirements for bridge defect identification.

[0033] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for identifying bridge defects based on low-altitude unmanned aerial vehicles (UAVs), characterized in that, Includes the following steps: S1: Obtain the architectural information model of the target bridge, and extract the three-dimensional spatial features and attribute features of the bridge components of the target bridge; S2: Based on the three-dimensional spatial features and attribute features, the target bridge is divided into a global detection area and a fine detection area, and corresponding inspection constraint parameters are configured for different detection areas; S3: Generate corresponding initial inspection routes based on the three-dimensional spatial features of the global detection area and the fine detection area; wherein, a fine inspection route including edge-following flight path and multi-angle shooting nodes is generated for the fine detection area. S4: Control the low-altitude UAV to fly along the initial inspection route and acquire real-time environmental perception data during the flight; S5: Spatially fuse and compare the real-time environmental perception data with the building information model. When the pre-set trigger adjustment conditions are met, the initial inspection route is dynamically adjusted under the premise of prioritizing the inspection constraint parameters. S6: Fly along the adjusted inspection route and collect bridge image data for bridge defect identification.

2. The bridge defect identification method based on low-altitude unmanned aerial vehicles according to claim 1, characterized in that, The extraction of the three-dimensional spatial features and attribute features of the bridge components included in the target bridge includes: The building information model is simplified geometrically, and non-core detailed components are removed. Extract the three-dimensional contour coordinates, spatial distribution relationship, and disease susceptibility level of the retained bridge components, and establish the mapping relationship between component type and the three-dimensional spatial features and attribute features.

3. The bridge defect identification method based on low-altitude unmanned aerial vehicles according to claim 2, characterized in that, Based on the three-dimensional spatial features and attribute features, the target bridge is divided into a global detection region and a fine detection region, and corresponding inspection constraint parameters are configured for different detection regions, including: Identify exposed and regularly shaped bridge components in the building information model, and divide the spatial range of their location into the global detection area; Identify components in the building information model that are hidden, structurally irregular, and have a susceptibility level to defects higher than a preset threshold, and divide the spatial range of these components into the fine detection area. Based on the type and size of different components within the fine detection area, the inspection constraint parameters are generated. The inspection constraint parameters include at least: target shooting distance range, shooting angle coverage range, and flight speed limit.

4. The bridge defect identification method based on low-altitude unmanned aerial vehicles according to claim 3, characterized in that, The step of generating corresponding initial inspection routes based on the three-dimensional spatial features of the global detection area and the fine detection area includes: For the global detection area, a global inspection route is generated based on the outer contour of the main structure of the target bridge, which is used to fly around the target bridge and satisfy the preset heading and lateral overlap rate. For the fine inspection area, based on the three-dimensional contour coordinates of the target component, a path trajectory parallel to the outer surface of the component is extracted to generate an edge-following flight path whose distance from the component surface is within the target shooting distance range; according to the three-dimensional structural morphology characteristics of the target bridge component in the fine inspection area, multiple key shooting angle nodes that meet the shooting angle coverage range are set on the edge-following flight path to generate the fine inspection route.

5. The bridge defect identification method based on low-altitude unmanned aerial vehicles according to claim 1, characterized in that, The step of spatially fusing and comparing the real-time environmental perception data with the building information model, and dynamically adjusting the initial inspection route when preset trigger adjustment conditions are met, includes: If the real-time environmental perception data indicates that there is an obstacle ahead of the flight path and the distance to the initial inspection route is less than the preset safe distance, then the obstacle avoidance adjustment condition is triggered; a local detour path is planned based on the pathfinding algorithm, and the inspection constraint parameters are used as the forced constraint nodes of the pathfinding algorithm, so that the local detour path can avoid obstacles while satisfying the set edge distance and shooting angle. If the real-time environmental perception data indicates that the displacement deviation between the actual spatial position of the bridge component and the three-dimensional spatial coordinates in the building information model is greater than or equal to a preset deviation threshold, then the registration adjustment condition is triggered; the real-time point cloud data is spatially registered with the building information model using a point cloud registration algorithm, and the waypoint coordinates in the initial inspection route are synchronously corrected based on the registration offset.

6. The bridge defect identification method based on low-altitude unmanned aerial vehicles according to claim 1, characterized in that, The process of flying along the adjusted inspection route and collecting bridge image data includes: When the low-altitude UAV is in the global detection area, it triggers shooting at equal intervals according to the first preset flight speed; When the low-altitude UAV is in the fine detection area, it flies at a second preset flight speed lower than the first preset flight speed, and when it reaches the multi-angle shooting node set in the fine inspection route, it controls the camera attitude and triggers shooting at the second preset shooting frequency.

7. The bridge defect identification method based on low-altitude unmanned aerial vehicles according to claim 1, characterized in that, The acquisition of real-time environmental perception data includes: Simultaneously acquire depth point cloud data from airborne lidar and image pixel data from visual sensors; The depth point cloud data and the image pixel data are time-stamp aligned and spatial coordinate calibrated. The calibrated data is fused to construct a real-time 3D point cloud map, and the 3D boundary contours of potential obstacles are extracted based on the real-time 3D point cloud map.

8. The bridge defect identification method based on low-altitude unmanned aerial vehicles according to claim 5, characterized in that, The dynamic adjustment of the initial inspection route also includes anomaly handling steps: When the pathfinding algorithm determines that a local detour path cannot be generated under the constraints of the target shooting distance range and the shooting angle coverage range, the low-altitude UAV is controlled to hover and the current route interruption point is recorded. A reverse safe evacuation trajectory is generated based on the location of the flight route interruption, and the bridge components corresponding to the location of the flight route interruption are marked as requiring manual re-inspection in the building information model.

9. The bridge defect identification method based on low-altitude unmanned aerial vehicles according to claim 1, characterized in that, The bridge defect identification process includes: Extract two-dimensional features of bridge defects from the collected bridge image data; Based on the spatial pose data of the low-altitude UAV and the intrinsic parameter matrix of the camera when collecting the bridge image data, the two-dimensional features of the defects are mapped back to the three-dimensional spatial coordinate system of the building information model. The building information model automatically generates label information including the type of disease, the three-dimensional coordinates of the disease, and the size of the disease.

Images