A drone detection and localization system and method

Abstract

This application primarily relates to the field of UAV detection and positioning technology at docks, specifically to a UAV detection and positioning system and method. The UAV detection and positioning system provided in this application includes: a satellite positioning module for acquiring the UAV's original global positioning coordinates; a structure perception module integrated on the UAV for collecting data on the steel structure to be inspected; a data fusion module communicatively connected to both the satellite positioning module and the structure perception module for generating corrected positioning coordinates; and a flight control module for generating flight control commands based on the corrected positioning coordinates to guide the UAV in inspecting the steel structure. Ultimately, this achieves stable, high-precision absolute positioning of the UAV in complex environments.

Description

Technical Field

[0001] This application mainly relates to the field of drone detection and positioning technology at docks, specifically to a drone detection and positioning system and method. Background Technology

[0002] Large steel structures such as quay cranes at ports and wharves operate in harsh environments with high intensity, high load, and high salt spray for extended periods, making their metal structures highly susceptible to progressive damage such as cracks, corrosion, and loose bolts. Failure to promptly inspect and maintain these structures can lead to serious safety accidents. Traditional inspection methods often rely on manual visual inspection or the use of aerial work platforms, which suffer from numerous blind spots, high-risk high-altitude operations, and low efficiency.

[0003] With the development of drone technology, automated inspection using drones has become a trend. Current technologies typically plan the drone's flight path based on the equipment's 3D design model, and utilize technologies such as RTK (Real-Time Kinematic) for GPS positioning, enabling the drone to fly to pre-set inspection points and acquire images. However, this method has inherent drawbacks: First, GPS signals are easily interfered with by obstruction and reflection near large metal structures, resulting in decreased positioning accuracy. Second, there are deviations such as installation errors and structural deformations between the 3D model and the actual object, causing the location where the UAV flies to based on the model coordinates to differ from the actual location of the point to be inspected, making it difficult to achieve high-precision close-range inspection.

[0004] Therefore, how to achieve high-precision and high-reliability positioning of large steel structures by UAVs under conditions of unstable GPS signals and model-to-physical deviation is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] This application provides a UAV detection and positioning system and method, which solves the technical problem of high-precision positioning and detection of large steel structures under conditions of unstable GPS signals and model-to-object deviation.

[0006] This application proposes a UAV detection and positioning system to solve the above-mentioned technical problems, including a satellite positioning module, a structure perception module, a data fusion module, and a flight control module: The satellite positioning module is used to obtain the original global positioning coordinates of the UAV; The structure perception module is integrated on the UAV and is used to collect data on the steel structure to be inspected. The data fusion module is communicatively connected to the satellite positioning module and the structure perception module, respectively, and is used to match the data of the steel structure to be detected with the data in the preset three-dimensional model of the steel structure to determine the actual relative position of the UAV with respect to the steel structure to be detected. Based on the actual relative position and the theoretical target position determined by the preset three-dimensional steel structure model, the error vector is calculated; The original global positioning coordinates are corrected based on the error vector to generate corrected positioning coordinates; The flight control module is used to generate flight control commands based on the corrected positioning coordinates to guide the UAV to inspect the steel structure to be inspected.

[0007] In one embodiment of this application, the structure-aware module includes: A lidar unit is used to collect point cloud data of the steel structure to be inspected. The multi-view vision unit is used to acquire image data of the steel structure to be inspected and obtain local structural depth information.

[0008] In one embodiment of this application, the data fusion module further includes: Two-dimensional visual feature points are extracted from the image data, and the two-dimensional visual feature points are mapped to the corresponding point cloud feature points in the point cloud data. The point cloud feature points are matched with the preset three-dimensional steel structure model to obtain the rigid body transformation matrix of the point cloud coordinate system relative to the preset three-dimensional steel structure model coordinate system.

[0009] In one embodiment of this application, the data fusion module further includes: The iterative nearest point algorithm is used to iteratively match the feature points of the point cloud with the preset features of the three-dimensional steel structure model to obtain the rigid body transformation matrix.

[0010] In one embodiment of this application, the data fusion module further includes: Visual feature points are extracted from the image data using a scale-invariant feature transformation algorithm and / or an accelerated robust feature algorithm.

[0011] In one embodiment of this application, the corrected positioning coordinates are: P gpscorrected =P gpsraw +K (P targetmodel -P dronereal ); Among them, P gpsraw The original global positioning coordinates are given, K is the correction factor, and P is the original global positioning coordinates. targetmodel P is the theoretical target location determined for the pre-defined three-dimensional model of the steel structure. dronereal This represents the actual relative position.

[0012] In one embodiment of this application, the lidar unit is a surround-view lidar, used to perform a 360-degree scan of the environment surrounding the steel structure to be inspected, so as to generate point cloud data characterizing the spatial contour of the steel structure.

[0013] In one embodiment of this application, the data of the steel structure to be inspected includes at least one of the following: steel structure edges, corners, nodes, welds, and bolted connection points.

[0014] This application also provides a method for detecting and locating unmanned aerial vehicles (UAVs) in the aforementioned UAV detection and positioning system, the method comprising: Obtain the drone's original global positioning coordinates; Collect data from the steel structure to be inspected; The data of the steel structure to be inspected is matched with the data in the preset three-dimensional model of the steel structure to determine the actual relative position of the UAV with respect to the steel structure to be inspected. Based on the actual relative position and the theoretical target position determined by the preset three-dimensional steel structure model, the error vector is calculated; The original global positioning coordinates are corrected based on the error vector to generate corrected positioning coordinates; Flight control commands are generated based on the corrected positioning coordinates to guide the UAV to inspect the steel structure to be inspected.

[0015] In one embodiment of this application, the data of the steel structure to be detected includes: point cloud data of the steel structure to be detected and image data of the steel structure to be detected.

[0016] In one embodiment of this application, matching the data of the steel structure to be inspected with the data in a preset three-dimensional model of the steel structure to determine the actual relative position of the UAV with respect to the steel structure to be inspected further includes: Two-dimensional visual feature points are extracted from the image data, and the two-dimensional visual feature points are mapped to the corresponding point cloud feature points in the point cloud data. The point cloud feature points are matched with the preset three-dimensional steel structure model to obtain the rigid body transformation matrix of the point cloud coordinate system relative to the preset three-dimensional steel structure model coordinate system.

[0017] In one embodiment of this application, the point cloud feature points and the preset three-dimensional steel structure model features are iteratively matched using an iterative nearest-point algorithm to obtain the rigid body transformation matrix.

[0018] This application also provides a drone, including the drone detection and positioning system as described above.

[0019] This application provides a UAV detection and positioning system and method that achieves centimeter-level precise relative positioning of the UAV relative to a steel structure by fusing laser point cloud and multi-view visual data. This high-precision relative position information is then used as a true reference to perform real-time error calculation and automatic compensation correction on the satellite absolute positioning coordinates, ultimately achieving stable and high-precision absolute positioning of the UAV in complex environments. Attached Figure Description

[0020] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the specific embodiments of this application will be described in detail below with reference to the accompanying drawings, wherein: Figure 1 A structural diagram of an embodiment of the UAV detection and positioning system of this application is disclosed; Figure 2 A flowchart of a drone detection and positioning method according to an embodiment of this application is disclosed. Detailed Implementation

[0021] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the specific embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0022] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0023] Furthermore, the terms “up,” “down,” “left,” “right,” “top,” “bottom,” “horizontal,” and “vertical” used in the following description should be understood as the orientations shown in the paragraph and related figures. This relative terminology is for illustrative purposes only and does not imply that the described device must be manufactured or operated in a specific orientation, and therefore should not be construed as a limitation of this application.

[0024] It is understood that although terms such as "first," "second," and "third" may be used herein to describe various components, regions, layers, and / or parts, these components, regions, layers, and / or parts should not be limited by these terms, and these terms are only used to distinguish different components, regions, layers, and / or parts. Therefore, the first component, region, layer, and / or part discussed below may be referred to as the second component, region, layer, and / or part without departing from some embodiments of this application.

[0025] Figure 1A structural diagram of an embodiment of the UAV detection and positioning system of this application is disclosed, as follows: Figure 1 As shown, this application provides a UAV detection and positioning system, including a satellite positioning module 10, a structure perception module 20, a data fusion module 30, and a flight control module 40: Satellite positioning module 10 is used to obtain the raw global positioning coordinates of the UAV; The structural sensing module 20, integrated on the UAV, is used to collect data on the steel structure to be inspected; The data fusion module 30 is communicatively connected to the satellite positioning module 10 and the structure perception module 20, respectively, and is used to match the data of the steel structure to be detected with the data in the preset three-dimensional model of the steel structure to determine the actual relative position of the UAV with respect to the steel structure to be detected. Based on the actual relative position and the theoretical target position determined by the preset three-dimensional steel structure model, the error vector is calculated; The original global positioning coordinates are corrected based on the error vector to generate corrected positioning coordinates; The flight control module 40 is used to generate flight control commands based on the corrected positioning coordinates to guide the UAV to inspect the steel structure to be inspected.

[0026] This application provides a UAV detection and positioning system, which is particularly suitable for detection scenarios of large steel structures such as port terminals, quay cranes, and mobile port machinery. By integrating satellite global positioning and relative positioning based on the characteristics of the steel structure, a redundant positioning system is constructed. Using the actual three-dimensional geometric features of the steel structure as the absolute reference, the system achieves high-precision positioning of the UAV relative to the structure. The positioning result is used as a dynamic correction quantity to compensate for the satellite main positioning information in real time, thereby enabling the UAV to achieve autonomous positioning and automatic compensation for precise flight in complex environments. Even in the case of severe satellite signal obstruction or interference, it can still achieve stable positioning by relying on the physical structural features.

[0027] The following is a detailed description of each module of the UAV detection and positioning system proposed in this application. It should be understood that, within the scope of this application, the above-mentioned technical features of this application and the technical features specifically described below (such as in the embodiments) can be combined and related to each other to form a preferred technical solution.

[0028] Satellite positioning module 10 is used to obtain the raw global positioning coordinates of the UAV.

[0029] The satellite positioning module 10 may include GPS (Global Positioning System), BeiDou and other satellite navigation systems. The following embodiments use GPS as an example for illustration.

[0030] Specifically, the GPS positioning module can provide global coordinates for RTK (Real-Time Kinematic) / PPP (Precise Point Positioning) receivers.

[0031] During the takeoff and long-range flight phases of the UAV, the system uses global coordinates provided by GPS and RTK for route planning and flight control, serving as the primary basis for UAV route planning and long-distance navigation.

[0032] RTK is a high-precision GPS positioning enhancement technology that can improve positioning accuracy from meters to centimeters through differential correction between ground reference stations and rover stations.

[0033] PPP uses a single GNSS (Global Navigation Satellite System) receiver, which utilizes precise ephemeris and satellite clock bias provided by the International GNSS Service (IGS) to achieve high-precision positioning at the millimeter to decimeter level based on carrier phase observations.

[0034] However, in scenarios such as dock operation areas, large steel structures such as quay cranes and gantry cranes are prone to blocking or multipath interference with GPS signals. Simply relying on RTK / PPP is insufficient to meet the accurate positioning requirements of detection points. Therefore, it is necessary to combine subsequent structural perception and data fusion solutions to achieve error compensation.

[0035] The structural perception module 20, integrated on the UAV, is used to collect data on the steel structure to be inspected. Furthermore, the steel structure to be inspected is a large steel structure in a port terminal setting, such as the steel structure of a quay crane or a mobile port machinery equipment.

[0036] In one embodiment, the structure sensing module 20 further includes: A lidar unit is used to acquire point cloud data of the steel structure area to be measured. A multi-view vision unit is used to simultaneously acquire image data of the steel structure area to be detected.

[0037] This embodiment integrates lidar and multi-view cameras to achieve real-time error correction for the positioning of the steel structure of the port machinery at the dock, ensuring anti-interference and real-time performance in the port machinery environment.

[0038] In some embodiments, for key inspection areas such as the main beam welds of quay cranes, drum bolt connection points, and steel structure nodes of mobile port machinery, the lidar unit can prioritize the acquisition of high-resolution point cloud data, while the multi-view vision unit simultaneously captures the visual features of subtle defects such as weld cracks and loose bolts, thereby achieving integrated positioning and inspection.

[0039] A LiDAR (Light Detection and Ranging) unit acquires the precise three-dimensional coordinates of the surrounding object's surface by emitting a laser beam and measuring the time it takes for the reflected echo to travel. In this application, LiDAR can be used to extract edge feature points (such as welds and bolts) of a steel structure to generate point cloud data. The point cloud data is a collection of a large number of three-dimensional spatial points, each containing X, Y, and Z coordinate information, and sometimes also including attributes such as reflection intensity and color, used to describe the geometry of the object's surface.

[0040] In one embodiment, the lidar unit is a surround-view lidar used to perform a 360-degree scan of the environment surrounding the steel structure to be inspected, in order to generate point cloud data characterizing the spatial contour of the steel structure.

[0041] The three-dimensional model involved in this application specifically refers to the design model of steel structures (such as port machinery equipment), which is usually derived from BIM (Building Information Modeling) or CAD (Computer-Aided Design) software and records the theoretical geometric dimensions and feature point locations of the structure. In complex environments such as ports and docks, this model can provide a precise flight path reference for UAVs.

[0042] In conjunction with this, a multi-view vision unit can be a multi-view camera, whose core function is to provide stereo vision capabilities, acquire multi-view images of the target area, and use stereo vision algorithms to convert two-dimensional image feature points into local three-dimensional coordinates to obtain the precise spatial position of key structures such as welds and bolts. This not only improves the accuracy of depth calculation but also reduces blind spots caused by occlusion areas, and can also cross-validate the measurement results.

[0043] Stereo vision is a technique that recovers the depth information of a three-dimensional scene by analyzing the disparity between the left and right images acquired by binocular (multi-)cameras. Its core task is stereo matching, which is to estimate the disparity value of each pixel.

[0044] Addressing the challenges of large-scale, complex steel structures in port terminals, and the high-altitude, confined spaces of critical inspection areas (such as welds on quay crane main beams and bolt connection points), this application employs a multi-view camera mounted on a drone. This complements the LiDAR system, creating a "vision-laser" system. LiDAR excels at acquiring large-scale structural frame points, adapting to the geometric feature clouds of steel structures extending across a wide spatial area. Multi-view vision, on the other hand, excels at capturing high-precision local 3D information (depth information) of small structures like welds and bolts, compensating for LiDAR's limitations in identifying minute structures. The generated point cloud data is then compared with a 3D model to determine the frame of the steel structure to be inspected. This combination enables this application to acquire both the overall structural outline and accurately locate minute inspection targets in complex steel structure scenarios such as port machinery.

[0045] The data fusion module 30 is communicatively connected to the satellite positioning module 10 and the structure perception module 20 respectively. The data of the steel structure to be detected is matched with the data in the preset three-dimensional model of the steel structure to determine the actual relative position of the UAV with respect to the steel structure to be detected. Based on the actual relative position and the theoretical target position determined by the preset three-dimensional steel structure model, the error vector is calculated; The original global positioning coordinates are corrected based on the error vector to generate corrected positioning coordinates.

[0046] Specifically, the data fusion module 30 can use an edge computing chip with a built-in iterative nearest point algorithm. It can use the detected relative position information of the steel structure to fine-tune the original global positioning coordinates and generate a precise positioning command for the UAV relative to the steel structure, so as to enable the UAV to accurately approach the detection point.

[0047] In one embodiment, the data fusion module 30 further includes: Two-dimensional visual feature points are extracted from the image data, and the two-dimensional visual feature points are mapped to the corresponding point cloud feature points in the point cloud data. The point cloud feature points are matched with the preset three-dimensional steel structure model to obtain the rigid body transformation matrix of the point cloud coordinate system relative to the preset three-dimensional steel structure model coordinate system.

[0048] To achieve coordinate matching and error calculation, the data fusion module 30 performs the following specific processing flow: First, feature extraction is performed on the point cloud data and the image data. Specifically, two-dimensional visual feature points are extracted from the image data, and these two-dimensional visual feature points are mapped to the corresponding point cloud feature points in the point cloud data. For example, key feature points (such as steel structure nodes and welds) are extracted from the point cloud data, and two-dimensional feature points (such as edges and corners) of the target are extracted from the image data using visual algorithms.

[0049] Optionally, visual feature points can be extracted from image data using the SIFT and / or SURF algorithms. SIFT (Scale-Invariant Feature Transform) is a computer vision feature extraction algorithm that can detect and describe local feature points (such as corners and edges) in an image, and is invariant to image scaling, rotation, and illumination changes. SURF (Speeded-Up Robust Features) extracts scale-invariant features and is a robust image recognition and description algorithm that is more efficient than SIFT.

[0050] By utilizing multi-view cameras to provide stereo vision capabilities, multi-view images of the target area are acquired. SIFT / SURF algorithms are then employed to extract feature points from the images, achieving stable and reliable image feature matching. Furthermore, by leveraging the principle of parallax, two-dimensional image feature points are converted into local three-dimensional coordinates, allowing for the acquisition of precise spatial locations of key structures such as welds and bolts. This provides high-precision local geometric information for subsequent coordinate matching and error calculation.

[0051] Based on this, the point cloud feature points are matched with the preset three-dimensional steel structure model to obtain the rigid body transformation matrix of the point cloud coordinate system relative to the preset three-dimensional steel structure model coordinate system.

[0052] In one embodiment, the data fusion module further includes: The iterative nearest point algorithm is used to iteratively match the feature points of the point cloud with the preset features of the three-dimensional steel structure model to obtain the rigid body transformation matrix.

[0053] In this embodiment, the point cloud features P are determined based on the ICP (Iterative Closest Point) algorithm. L With 3D model features P M The matching process is performed to calculate the rigid body transformation matrix of the point cloud coordinate system relative to the 3D model coordinate system.

[0054] Among them, the ICP algorithm is a point-to-point registration method used to calculate the optimal rotation and translation transformation between two point cloud datasets to achieve accurate alignment. Based on the least squares method, this algorithm iteratively optimizes the transformation parameters through two steps: "finding the nearest corresponding point" and "solving the optimal transformation," until the mean square error is less than a set threshold.

[0055] A rigid body transformation matrix is ​​a mathematical matrix that describes the changes in position and orientation of an object in three-dimensional space. It contains rotational and translational components and is used to transform a point in one coordinate system to another.

[0056] This application uses the ICP algorithm to iterate the measured point cloud feature points and the preset point cloud repeatedly to obtain an accurate rigid body transformation matrix. Then, the UAV system establishes an accurate mapping relationship between the "point cloud coordinate system" and the "model coordinate system".

[0057] This application first uses the initial coordinates provided by GPS as the basis for coarse alignment to narrow the search range. It then uses feature point downsampling to extract key features such as corner points and edge points for iterative optimization of registration. The UAV onboard coordinates are mapped to the model coordinate system through a transformation matrix. Considering the steel structure characteristics of port machinery, feature point downsampling can prioritize the retention of key positioning features such as beam-column connection nodes and welds, thereby improving matching efficiency and accuracy.

[0058] The data fusion module calculates the theoretical coordinates P of the point to be detected based on a preset 3D model of the steel structure. targetmodel Then, based on the point cloud matching results above, obtain the actual coordinates P of the UAV relative to the steel structure to be inspected. dronereal .

[0059] Based on this, the deviation between the UAV's original global positioning coordinates and its actual coordinates is calculated using the following formula: e error = P targetmodel -P dronereal ).

[0060] Data fusion module 30 based on error vector e error The GPS coordinates were fine-tuned to obtain the corrected, precise coordinates P. gpscorrected The corrected positioning coordinates are calculated using the following formula: P gpscorrected =P gpsraw +K (P targetmodel -P dronereal ); Among them, P gpsraw The original global positioning coordinates are given, K is the correction factor, and P is the original global positioning coordinates. targetmodel P is the theoretical target location determined for the pre-defined three-dimensional model of the steel structure. dronereal This represents the actual relative position.

[0061] The flight control module 40 is used to generate flight control commands based on the corrected positioning coordinates, and control the motors to guide the UAV to perform high-precision inspection of the steel structure to be inspected.

[0062] Based on the aforementioned UAV detection and positioning system, this application also provides a corresponding UAV detection and positioning method.

[0063] This application provides a method for detecting and locating unmanned aerial vehicles (UAVs), which is applied to UAVs. The method includes: Obtain the drone's original global positioning coordinates; Collect data from the steel structure to be inspected; The data of the steel structure to be inspected is matched with the data in the preset three-dimensional model of the steel structure to determine the actual relative position of the UAV with respect to the steel structure to be inspected. Based on the actual relative position and the theoretical target position determined by the preset three-dimensional steel structure model, the error vector is calculated; The original global positioning coordinates are corrected based on the error vector to generate corrected positioning coordinates; Flight control commands are generated based on the corrected positioning coordinates to guide the UAV to inspect the steel structure to be inspected.

[0064] Flowcharts are used in this application to illustrate the operations performed by the system according to embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, various steps can be processed in reverse order or simultaneously. Furthermore, other operations may be added to these processes, or one or more steps may be removed from these processes.

[0065] Figure 2 A flowchart of a drone detection and positioning method according to an embodiment of this application is disclosed, as follows: Figure 2 As shown, the specific execution process of the UAV detection and positioning method provided in this application is as follows: First, the initial parameter configuration is completed, and the overall operation route is planned. After the UAV takes off, the flight control module reads the port map and the preset three-dimensional model data of the steel structure to be inspected according to the preset detection task, combines it with GPS positioning information, and flies to the area of ​​the steel structure to be inspected according to the planned route.

[0066] Once the drone has determined that it has entered the predetermined area of ​​the steel structure using GPS coordinates, it collects data on the steel structure to be inspected.

[0067] In one embodiment, the data of the steel structure to be detected includes: point cloud data of the steel structure to be detected and image data of the steel structure to be detected.

[0068] Specifically, the LiDAR device begins a 360-degree rotating scan, generating 3D point cloud data of the surrounding environment, which is then transmitted to the data fusion module. Simultaneously, multiple cameras begin synchronously acquiring image frames and calculating the depth information of each pixel using a stereo matching algorithm to generate a local depth map.

[0069] In one embodiment, two-dimensional visual feature points are extracted from the image data, and the two-dimensional visual feature points are mapped to the point cloud feature points corresponding to the point cloud data; The point cloud feature points are matched with the preset three-dimensional steel structure model to obtain the rigid body transformation matrix of the point cloud coordinate system relative to the preset three-dimensional steel structure model coordinate system.

[0070] In this embodiment, after receiving the point cloud data, filtering and downsampling are first performed to remove outliers and retain structural edge features. Then, point cloud feature extraction algorithms (such as SIFT and / or SURF algorithms) are used to extract significant feature points of the steel structure, such as corner points, beam-column connection nodes, and weld areas.

[0071] In one embodiment, an iterative nearest-point algorithm is used to iteratively match the feature points of the point cloud with the preset features of the three-dimensional steel structure model to obtain the rigid body transformation matrix.

[0072] By calling a pre-defined 3D steel structure model, corresponding feature points are extracted from the model. The ICP (Iterative Closest Point) algorithm is used to register the feature points in the point cloud with those in the model. The rigid body transformation matrix is ​​iteratively solved to achieve optimal alignment between the feature points in the point cloud coordinate system and those in the model coordinate system. Through this step, a precise mapping relationship between the point cloud and the model is established.

[0073] Assume the theoretical coordinates of the point to be detected in the pre-defined 3D model of the steel structure are P. targetmodel (For example, the coordinates of a bolt connection point that needs to be photographed). Based on the rigid body transformation matrix obtained from the above steps, the model coordinates of the point to be measured can be mapped to the current point cloud coordinate system, and further transformed to the UAV body coordinate system, thus obtaining the UAV's ideal flight position relative to the target point. However, in actual flight, due to GPS errors, attitude drift, and other factors, the actual position P of the UAV may vary. dronereal There is a deviation from the theoretical position. In the high-altitude inspection scenario of quay cranes, this deviation directly affects the accuracy of identifying defects such as weld cracks and loose bolts. The error correction scheme in this application can effectively solve this problem.

[0074] Subsequently, the error vector e is calculated. error The error vector directly reflects the deviation between the "theoretical position of the model" and the "actual position of the current UAV relative to the physical object".

[0075] According to the error vector e error The GPS coordinates were fine-tuned to obtain the corrected, precise coordinates P. gpscorrected The corrected positioning coordinates are calculated using the following formula: P gpscorrected =P gpsraw +K (P targetmodel -P dronereal ); Among them, Pgpsraw Let K be the correction factor, and P be the correction factor. targetmodel P is the theoretical target location determined for the pre-defined three-dimensional model of the steel structure. dronereal This represents the actual relative position.

[0076] Finally, based on the received corrected coordinates P dronereal As the new target location, flight control commands are generated to adjust the drone's attitude and position, enabling it to accurately approach the point to be detected.

[0077] When the UAV approaches the steel structure, it initiates point cloud and visual fusion positioning, calculating the error vector in real time and correcting the GPS coordinates. In environments such as ports and docks where numerous metal structures obstruct and reflect GPS signals, this solution effectively suppresses positioning drift caused by multipath effects. This significantly improves image acquisition quality, clearly capturing weld and bolt details. Furthermore, even in areas within the steel structure where GPS signals are completely lost, the system can still maintain relative positioning using LiDAR and visual features to complete subsequent inspection tasks, demonstrating superior anti-interference capabilities and robustness.

[0078] The UAV detection and positioning system and method provided in this application have the following beneficial effects: 1) By integrating GPS global positioning with relative positioning based on steel structure features, a redundant positioning system is constructed, which can still achieve stable positioning by relying on physical structure features even in port operation environments where GPS signals are blocked or severely interfered with. 2) By using lidar point cloud and multi-view vision to acquire the actual three-dimensional geometric information of the steel structure in real time and matching it with the model, it can effectively compensate for deviations caused by factors such as installation errors and structural deformation, achieve centimeter-level high-precision positioning, and meet the detection accuracy requirements of key parts such as quay crane welds and port machinery bolts. 3) The error vector calculated based on structural features is fed back to the GPS positioning information in real time, forming a two-layer closed-loop control to ensure that the UAV can accurately approach the target detection point, thereby improving the reliability of the inspection and the quality of image acquisition.

[0079] This application also provides an unmanned aerial vehicle (UAV), including the UAV detection and positioning system described above. During actual flight operations, the UAV is deeply coupled with the aforementioned detection and positioning system via an onboard flight control module, forming a complete closed-loop control link. The UAV provided in this application, by integrating the aforementioned UAV detection and positioning system, can achieve high-precision and high-reliability autonomous positioning and flight control in complex operating environments. Especially in scenarios such as port terminals, quay cranes, and mobile port machinery with densely distributed large steel structures, this UAV can effectively cope with interference factors such as GPS signal blockage, multipath effects, and strong reflections from metal structures. Relying on the fusion perception of lidar and multi-view vision, it matches the three-dimensional model of the steel structure in real time and dynamically corrects positioning errors, thereby ensuring accurate execution of the detection path and high-quality data acquisition at key nodes.

[0080] The basic concepts have been described above. Obviously, for those skilled in the art, the above disclosure is merely illustrative and does not constitute a limitation of this application. Although not explicitly stated herein, those skilled in the art may make various modifications, improvements, and corrections to this application. Such modifications, improvements, and corrections are suggested in this application, and therefore remain within the spirit and scope of the exemplary embodiments of this application.

[0081] Furthermore, this application uses specific terms to describe embodiments of the application. For example, "an embodiment," "one embodiment," and / or "some embodiments" refer to a particular feature, structure, or characteristic related to at least one embodiment of the application. Therefore, it should be emphasized and noted that "an embodiment," "one embodiment," or "an alternative embodiment" mentioned twice or more in different locations in this specification do not necessarily refer to the same embodiment. In addition, certain features, structures, or characteristics in one or more embodiments of the application can be appropriately combined.

[0082] In some embodiments, numbers describing the quantity of components and attributes are used. It should be understood that such numbers used to describe embodiments are sometimes modified by the terms "approximately," "approximately," or "generally." Unless otherwise stated, "approximately," "approximately," or "generally" indicates that the numbers are allowed to vary by ±20%. Accordingly, in some embodiments, the numerical parameters used in this application are approximate values, which may be changed depending on the characteristics required by individual embodiments. In some embodiments, numerical parameters should take into account specified significant digits and employ a general method of digit reservation. Although the numerical ranges and parameters used to confirm their breadth of range in some embodiments of this application are approximate values, in specific embodiments, such values ​​are set as precisely as feasible.

Claims

1. A drone detection and positioning system, characterized in that, It includes a satellite positioning module, a structure perception module, a data fusion module, and a flight control module: The satellite positioning module is used to obtain the original global positioning coordinates of the UAV; The structure perception module is integrated on the UAV and is used to collect data on the steel structure to be inspected. The data fusion module is communicatively connected to the satellite positioning module and the structure perception module, respectively, and is used to match the data of the steel structure to be detected with the data in the preset three-dimensional model of the steel structure to determine the actual relative position of the UAV with respect to the steel structure to be detected. Based on the actual relative position and the theoretical target position determined by the preset three-dimensional steel structure model, the error vector is calculated; The original global positioning coordinates are corrected based on the error vector to generate corrected positioning coordinates; The flight control module is used to generate flight control commands based on the corrected positioning coordinates to guide the UAV to inspect the steel structure to be inspected.

2. The UAV detection and positioning system as described in claim 1, characterized in that, The structure sensing module includes: A lidar unit is used to collect point cloud data of the steel structure to be inspected. The multi-view vision unit is used to acquire image data of the steel structure to be inspected and obtain local structural depth information.

3. The UAV detection and positioning system as described in claim 2, characterized in that, The data fusion module further includes: Two-dimensional visual feature points are extracted from the image data, and the two-dimensional visual feature points are mapped to the corresponding point cloud feature points in the point cloud data. The point cloud feature points are matched with the preset three-dimensional steel structure model to obtain the rigid body transformation matrix of the point cloud coordinate system relative to the preset three-dimensional steel structure model coordinate system.

4. The UAV detection and positioning system as described in claim 3, characterized in that, The data fusion module further includes: The iterative nearest point algorithm is used to iteratively match the feature points of the point cloud with the preset features of the three-dimensional steel structure model to obtain the rigid body transformation matrix.

5. The UAV detection and positioning system as described in claim 3, characterized in that, The data fusion module further includes: Visual feature points are extracted from the image data using a scale-invariant feature transformation algorithm and / or an accelerated robust feature algorithm.

6. The UAV detection and positioning system as described in claim 1, characterized in that, The corrected positioning coordinates are: P gpscorrected =P gpsraw +K (P targetmodel -P dronereal ); Among them, P gpsraw The original global positioning coordinates are given, K is the correction factor, and P is the original global positioning coordinates. targetmodel P is the theoretical target location determined for the pre-defined three-dimensional model of the steel structure. dronereal This represents the actual relative position.

7. The UAV detection and positioning system as described in claim 2, characterized in that, The lidar unit is a surround-view lidar, used to perform a 360-degree scan of the environment surrounding the steel structure to be inspected, in order to generate point cloud data characterizing the spatial contour of the steel structure.

8. The UAV detection and positioning system as described in claim 1, characterized in that, The data of the steel structure to be tested includes at least one of the following: steel structure edges, corners, nodes, welds, and bolt connections.

9. A method for detecting and locating unmanned aerial vehicles (UAVs), applied to the UAV detection and location system according to any one of claims 1-8, characterized in that, The method includes: Obtain the raw global positioning coordinates of the drone; Collect data from the steel structure to be inspected; The data of the steel structure to be inspected is matched with the data in the preset three-dimensional model of the steel structure to determine the actual relative position of the UAV with respect to the steel structure to be inspected. Based on the actual relative position and the theoretical target position determined by the preset three-dimensional steel structure model, the error vector is calculated; The original global positioning coordinates are corrected based on the error vector to generate corrected positioning coordinates; Flight control commands are generated based on the corrected positioning coordinates to guide the UAV to inspect the steel structure to be inspected.

10. The UAV detection and positioning method as described in claim 9, characterized in that, The data of the steel structure to be inspected includes: point cloud data of the steel structure to be inspected and image data of the steel structure to be inspected.

11. The UAV detection and positioning method as described in claim 10, characterized in that, The step of matching the data of the steel structure to be inspected with the data in the preset three-dimensional model of the steel structure to determine the actual relative position of the UAV with respect to the steel structure to be inspected further includes: Two-dimensional visual feature points are extracted from the image data, and the two-dimensional visual feature points are mapped to the corresponding point cloud feature points in the point cloud data. The point cloud feature points are matched with the preset three-dimensional steel structure model to obtain the rigid body transformation matrix of the point cloud coordinate system relative to the preset three-dimensional steel structure model coordinate system.

12. The UAV detection and positioning method as described in claim 11, characterized in that, The point cloud feature points are iteratively matched with the preset steel structure 3D model features using an iterative nearest-point algorithm to obtain the rigid body transformation matrix.

13. A drone, characterized in that, Including the drone detection and positioning system as described in any one of claims 1 to 8.

Images