A double-curvature arch dam inspection method based on unmanned aerial vehicle autonomous navigation technology

Abstract

The application provides a double-curvature arch dam inspection method based on an unmanned aerial vehicle autonomous navigation technology, and the specific implementation manner is as follows: S1, laser radar point cloud data acquisition, point cloud data filtering and point cloud data registration; S2, double-curvature fitting and double-curvature focal point estimation; S3, unmanned aerial vehicle navigation target point planning; S4, unmanned aerial vehicle navigation algorithm correction; and S5, double-curvature arch dam surface image acquisition. The application is based on an unmanned aerial vehicle system carrying a laser radar and a holder camera, can realize the autonomous navigation and autonomous image acquisition functions of the unmanned aerial vehicle on the surface of the double-curvature arch dam, and can improve the image acquisition efficiency of the unmanned aerial vehicle on the double-curvature arch dam.

Description

Technical Field

[0001] This invention relates to the field of hyperbolic arch dam inspection technology, and in particular to a hyperbolic arch dam inspection method based on UAV autonomous navigation technology. Background Technology

[0002] Hyperbolic arch dams are a common type of large dam, typically characterized by their height, large surface area, and thin thickness. The sheer height and vast area of ​​the arch dam's exterior make surface inspection extremely difficult. With the development of drone technology, drones are increasingly being used for dam inspection. However, because large hyperbolic arch dams are usually built in deep mountain valleys, the terrain and dam height often cause satellite-navigated drones to crash when flying downstream of the dam due to signal loss. This makes it difficult to conduct automated drone inspections effectively. Therefore, this invention proposes a hyperbolic arch dam inspection method based on drone autonomous navigation technology. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a method for inspecting hyperbolic arch dams based on UAV autonomous navigation technology. This method utilizes a lidar mounted on a UAV to collect point cloud data of the dam and its surrounding environment, thereby acquiring point cloud information of the surrounding environment. The point cloud data is then processed and calculated to identify the hyperbolic arch dam structure and calculate the foci of the hyperbola on horizontal planes at different heights. Using the foci of the hyperbola as reference points, the navigation target point for the UAV at the next moment is planned, enabling the UAV to autonomously navigate and fly along a predetermined path in front of the hyperbolic arch dam. Finally, a gimbal camera mounted on the UAV collects images of the hyperbolic arch dam surface, thus enabling autonomous navigation and inspection of the dam by the UAV. This method solves the problem of signal loss leading to drone crashes during automatic inspection of hyperbolic arch dams in existing technologies.

[0004] According to an embodiment of the present invention, a method for inspecting a hyperbolic arch dam based on unmanned aerial vehicle (UAV) autonomous navigation technology is specifically implemented as follows:

[0005] S1: LiDAR point cloud data acquisition, point cloud data filtering, and point cloud data registration.

[0006] S2: Hyperbola fitting and hyperbola focus estimation.

[0007] S3: Drone navigation target point planning.

[0008] S4: Drone navigation algorithm correction.

[0009] S5: Image acquisition of the surface of the hyperbolic arch dam.

[0010] Preferably, the lidar is horizontally mounted on the UAV, and the lidar can rotate around the vertical direction to obtain point cloud information of the hyperbolic arch dam and its surrounding environment.

[0011] Preferably, the point cloud data filtering refers to removing bad and invalid points caused by uneven reflections from the external environment and the surface of the hyperbolic arch dam during the scanning process.

[0012] Preferably, the point cloud data registration specifically involves using an iterative algorithm to search for and match nearby points, transforming the coordinates of the point cloud data from multiple scans to ensure they are in the same coordinate system, thereby obtaining the overall point cloud data of the hyperbolic arch dam and its surrounding environment.

[0013] Preferably, the hyperbola fitting and hyperbola focus estimation refer to: processing the overall point cloud data of the obtained hyperbola arch dam and its surrounding environment, retaining only the point cloud data within the hyperbola arch dam area; cutting and dividing the point cloud data according to different horizontal heights, and fitting the point cloud data with a hyperbola in the horizontal direction to find the optimal hyperbola; and then calculating the focus of the hyperbola.

[0014] Preferably, the UAV navigation target point refers to the navigation coordinates of the UAV at the next moment, which are in the same horizontal plane as the focus of the hyperbola and have a one-to-one mapping relationship with the focus.

[0015] Preferably, the UAV navigation target point planning refers to planning the UAV's flight target for the next moment, using the navigation target point of the next moment as the target.

[0016] Preferably, the UAV navigation algorithm correction refers to the process whereby, after the UAV reaches the current navigation target point, the UAV fits a hyperbola of the current horizontal plane using laser point cloud data and calculates the focus of the hyperbola. Based on this focus, the theoretical navigation target point of the UAV in the current horizontal plane can be calculated, and then the current actual position is corrected to obtain the final target position of the UAV in the current horizontal plane.

[0017] Preferably, the acquisition of the hyperbolic arch dam surface image refers to the UAV hovering after reaching the final target position in the current horizontal plane, and the gimbal camera scanning the dam surface image in the horizontal direction.

[0018] Preferably, after the image scanning is completed, the UAV flies to the next target point and hovers again to perform image scanning, ultimately achieving fully autonomous flight of the UAV and completing the image acquisition of the hyperbolic arch dam.

[0019] Compared to existing technologies, this invention offers the following advantages: Unmanned aerial vehicles (UAVs) can achieve autonomous navigation and automatic image acquisition on the surface of a hyperbolic arch dam using lidar navigation equipment, without the need for satellite navigation. Furthermore, this method allows UAVs to acquire images within the same horizontal plane while hovering, reducing flight distance and improving stability and image acquisition efficiency. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the horizontal plane scan of the UAV in an embodiment of the present invention.

[0021] Figure 2 This is a schematic diagram of point cloud scanning for UAV navigation target points in an embodiment of the present invention.

[0022] Figure 3 This is a schematic diagram of UAV image acquisition in an embodiment of the present invention. Detailed Implementation

[0023] The technical solutions of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0024] In the embodiments of the present invention, the sensor uses a multi-line lidar, which is horizontally mounted on the UAV and rotates to scan around the vertical direction, thereby obtaining point cloud information of the hyperbolic arch dam and its surrounding environment.

[0025] S1: LiDAR point cloud data acquisition, point cloud data filtering, and point cloud data registration;

[0026] When the drone flies to a certain height on the side of the hyperbolic arch dam at time t0, the lidar can acquire point cloud data information of the hyperbolic arch dam and its surrounding environment before that time.

[0027] Filtering the point cloud data removes bad and invalid points caused by uneven reflections from the external environment and the surface of the hyperbolic arch dam during the scanning process. Then, the point cloud data is registered by transforming the coordinates of the point cloud data from multiple scans to ensure they are in the same coordinate system, thus obtaining the overall point cloud data of the hyperbolic arch dam.

[0028] S2: Hyperbola fitting and hyperbola focus estimation;

[0029] As attached Figure 1 As shown, assuming the UAV is located at point P at time t0, the point cloud information in the horizontal plane where point P is located at that time includes the point cloud information of the hyperbolic arch dam and the slopes on both sides of the dam. This information cannot be directly fitted using the full point cloud information. Assume the angle between the UAV and the lines connecting the left and right sides of the hyperbolic arch dam is β (β < 180°). The point cloud information of the slopes on both sides of the hyperbolic arch dam can be filtered out using angle β, thus obtaining the point cloud information of the hyperbolic arch dam in the horizontal plane where point P is located.

[0030] Assume the focus of the hyperbola corresponding to the hyperbola is on the x-axis and the hyperbola lies on the positive x-axis. Then, the hyperbola fitting model extracts the expression from the aforementioned point cloud as follows: A point cloud hyperbola (where a0 > 0, b0 > 0) is fitted. Based on this fitted point cloud hyperbola, the coordinates of the foci of the hyperbola on the positive x-axis can be obtained. Since the above focal coordinates are calculated within the horizontal plane where point P is located, the corresponding spatial coordinates of this focal point are:

[0031] Since point P and focus C are on the same horizontal plane, and there exists a function... This allows point P to be mapped one-to-one with focus C. Therefore, the coordinates of point P, where the drone is located, are:

[0032]

[0033] S3: Unmanned Aerial Vehicle (UAV) navigation target point planning;

[0034] As attached Figure 2 As shown, assume point Q is the navigation target point of the UAV at the next moment, and assume the distance between the horizontal plane containing point Q and the horizontal plane containing point P is Δz. Since the point cloud data obtained by the multi-line lidar at time t0 and before includes the point cloud information of the horizontal plane containing point Q, the coordinates of point Q at time t0 can be calculated using the same method as in step S2:

[0035]

[0036] Therefore, point Q is the navigation target point planned by the drone in the next moment. Finally, this navigation target point is published to the drone flight control system.

[0037] S4: Drone navigation algorithm correction;

[0038] Once the UAV autonomously flies to target point Q according to the planned navigation target point, it can recalculate the theoretical point cloud hyperbola and hyperbola focus on the horizontal plane where the UAV is located at point Q at that moment, using the same method as step S2. It can also estimate the coordinates of the theoretical navigation target point Q' on the current horizontal plane where point Q is located. Then, the difference between the coordinates of Q and Q' is added as negative feedback to the navigation target point planning system, thereby correcting the navigation target point.

[0039] S5: Image acquisition of the surface of the hyperbolic arch dam;

[0040] After the UAV autonomously navigates to a target point H1, it hovers using laser point cloud feature information. At this point, the gimbal camera on board the UAV scans the dam surface from left to right (or in reverse) within the range defined by angle β, thereby acquiring image information of the dam surface. After image acquisition is complete, the UAV stops hovering and autonomously flies to the next navigation target point. When the UAV reaches target point H2, and the vertical height difference between the two points satisfies |H1-H2|>△H, the UAV hovers again and performs image acquisition in the same manner.

[0041] Furthermore, it should be noted that the specific embodiments described in this specification may be given different names, and the above description is merely illustrative of the structure of the present invention. Any minor or simple variations made based on the structure, features, and principles of the present invention are included within the scope of protection of the present invention. Those skilled in the art can make various modifications or additions to the described specific examples or adopt similar methods, as long as they do not deviate from the structure of the present invention or exceed the scope defined by the claims, all of which should fall within the scope of protection of the present invention.

Claims

1. A method for inspecting hyperbolic arch dams based on UAV autonomous navigation technology, characterized in that, Includes the following steps: S1. Use the lidar equipped on the drone to acquire point cloud data of the hyperbolic arch dam, and filter and register the acquired point cloud data to obtain the overall point cloud data of the hyperbolic arch dam and its surrounding environment. S2. Process the obtained overall point cloud data to obtain the optimal hyperbola fitting model corresponding to the hyperbola arch dam, and calculate the focus of the hyperbola fitting model. S3. Based on the obtained optimal hyperbola fitting model and focus, plan the UAV navigation target point and plan the UAV navigation target point for the next moment. S4. After the UAV reaches the previously planned navigation target point, the theoretical position of the UAV navigation algorithm and the current actual position of the UAV are corrected to obtain the final target position of the UAV in the current horizontal plane. S5. After the drone reaches the final target position, use the gimbal camera on the drone to collect images of the surface of the hyperbolic arch dam. Finally, follow the above steps to complete the fully autonomous flight of the drone and the image collection of the hyperbolic arch dam. The point cloud data processing in step S2 includes filtering effective point cloud data, hyperbolic fitting of point cloud data, and hyperbolic focus estimation; among which, filtering effective point cloud data refers to filtering the point cloud information in the horizontal plane where the UAV is located, eliminating point cloud data in the non-hyperbolic arch dam area, and retaining the point cloud data in the hyperbolic arch dam area as effective point cloud data. The method for point cloud hyperbola fitting and hyperbola focus estimation is as follows: extract the expression from the effective point cloud data. (where a0 > 0, b0 > 0) point cloud hyperbola, the optimal hyperbola is obtained, and based on the obtained point cloud hyperbola equation, the coordinates of the closest focus to the hyperbola are estimated, and the coordinates are: ( ,0, ); In step S3, the planning method for the UAV navigation target point is as follows: Define a unique function in any horizontal plane that is either higher or lower than the current horizontal plane of the drone: (1) Then, using the point cloud data of the UAV in the horizontal plane at the current moment, calculate its focal coordinates: ( ,0, Using the coordinates of the focus as the independent variable, and substituting them into function (1), the position of the point corresponding to the focus within the horizontal plane is estimated. , , This coordinate is the estimated target point for the drone's navigation; finally, this target point is sent to the drone's flight control system. The method for correcting the position of the UAV in step S4 is as follows: after the UAV flies to the navigation target point, the theoretical position of the UAV in the plane at that moment will be estimated again, and then the difference between the theoretical position and the actual coordinates of the UAV will be calculated. This difference will be added as negative feedback to the planning of the next navigation target point to complete the correction of the navigation target point.

2. The inspection method for hyperbolic arch dams based on UAV autonomous navigation technology according to claim 1, characterized in that: The lidar is horizontally arranged on the UAV and is vertically rotatable, allowing it to obtain point cloud information of the hyperbolic arch dam and its surrounding environment.

3. The inspection method for hyperbolic arch dams based on UAV autonomous navigation technology according to claim 1, characterized in that: The filtering of point cloud data in step S1 refers to removing bad and invalid points caused by uneven reflections from the external environment and wall surfaces during the LiDAR scanning process; the registration of point cloud data in step S1 refers to using an iterative algorithm to search for nearby points and match them, transforming the coordinates of the point cloud data from multiple scans to make them in the same coordinate system, thereby obtaining the overall point cloud data of the environment.

4. The method for inspecting a hyperbolic arch dam based on UAV autonomous navigation technology according to claim 1, characterized in that, The method for acquiring the surface image of the hyperbolic arch dam in step S5 is as follows: After the drone autonomously navigates to a target point, it hovers. At this time, the gimbal camera carried by the drone scans the surface of the hyperbolic arch dam in one direction within the width of the dam, acquiring image information of the dam surface. After completing the image information acquisition, the drone stops hovering and autonomously flies to the next navigation target point. Then, the drone calculates the height difference between the target point and the previous hovering point. When the height difference reaches a set value, the drone hovers again and performs image acquisition in the same way.

Images