Underground pipe gallery detection method based on unmanned aerial vehicle point cloud scanning

Abstract

The invention relates to the field of underground pipe gallery space detection, in particular to an underground pipe gallery detection method based on unmanned aerial vehicle point cloud scanning, and the method comprises the steps: 1, arranging a reflector plate, and collecting the three-dimensional coordinates of the reflector plate in an absolute coordinate system; 2, performing three-dimensional laser scanning when the unmanned aerial vehicle flies, performing point cloud acquisition at the entrance of the underground pipe gallery, acquiring spatial point cloud data by a laser radar carried by the unmanned aerial vehicle, and recording environment image information by a visual sensor; 3, establishing a relative coordinate system associated with spatial features, and carrying out point cloud SLAM calculation; step 4, calculating overground and underground continuous spatial point clouds based on an SLAM algorithm, and extracting three-dimensional coordinates in the scanning point clouds; 5, converting the point cloud in the relative coordinate system to an absolute coordinate system by using a Rodrigue matrix conversion method; and step 6, acquiring real spatial position information of the underground pipe gallery. The real spatial position of the underground pipe gallery in an absolute coordinate system is obtained through continuous scanning of the unmanned aerial vehicle in combination with ground GNSS measurement.

Description

Technical Field

[0001] The present invention relates to the field of underground pipe gallery space detection, and in particular to an underground pipe gallery detection method based on unmanned aerial vehicle point cloud scanning. Background Art

[0002] As an important part of the city's municipal infrastructure, underground pipe corridors undertake multiple functions such as water supply, drainage, communication, and electricity. They have a very important impact on the normal operation of the city and the quality of life of residents. The accurate mapping of their spatial position is of great significance for planning, maintenance, management, and transformation. However, with the acceleration of urbanization and the deepening of underground space development, the spatial layout of pipe corridors has become increasingly complex, especially for deep underground pipe corridors, which are usually in a very harsh environment. Not only is there insufficient light and poor air quality, but there may also be problems such as narrow structures and dense obstacles. For deep underground pipe corridors with harsh environments, traditional measurement methods are often unable to accurately and efficiently obtain the spatial information of the pipe corridors, which not only affects the daily operation and maintenance of the pipe corridors, but may also have an adverse impact on safety monitoring, hidden danger investigation, and disaster warning.

[0003] Therefore, how to break through the limitations of traditional technology and explore an efficient, accurate and low-risk detection method has become an urgent problem to be solved in the current field of underground utility corridor management. Summary of the invention

[0004] In view of the problems existing in the prior art in underground pipeline corridor detection, especially in the detection of underground pipeline corridors with great burial depth and harsh environment, it is difficult for traditional measurement methods to comprehensively and accurately obtain the real spatial position of the underground pipeline corridor. The purpose of the present invention is to provide an underground pipeline corridor detection method based on drone point cloud scanning, which uses a drone to obtain continuous spatial point cloud data of the ground and underground pipeline corridors, obtains the point cloud in a relative coordinate system through a SLAM algorithm, converts the point cloud to an absolute coordinate system using GNSS control points collected in advance on the ground, and finally draws the underground pipeline corridor plan and other drawings to obtain the real spatial position of the underground pipeline corridor.

[0005] To achieve the above object, the technical solution adopted by the present invention is: an underground pipe gallery detection method based on drone point cloud scanning, comprising the following steps: Step 1: Deploy ground reflectors and use a GNSS receiver in PTK mode to collect the three-dimensional coordinates of the reflectors in an absolute coordinate system; Step 2: Install a laser radar on the drone, connect the signal extension line to the remote control, and perform 3D laser scanning while the drone is flying to collect point cloud data at the entrance of the underground corridor. The scanning range covers all reflectors, and the terminal of the signal extension line is placed in the corridor. The drone is controlled to enter the underground corridor through the entrance. The laser radar carried by the drone collects spatial point cloud data, and the visual sensor records environmental image information. Step 3: Extract feature points in the environment through visual sensors, combine them with the point cloud features of the lidar, establish a relative coordinate system associated with spatial features, and perform point cloud SLAM solution based on state updates and observation updates; Step 4: Calculate the continuous spatial point cloud above and below the ground based on the SLAM algorithm, and use the difference in reflection intensity between the reflector and the surrounding environment in the point cloud to extract the three-dimensional coordinates of the reflector in the scanned point cloud in the relative coordinate system; Step 5: Determine the transformation parameters and transformation process, and apply the Rodrigues matrix transformation method to transform the point cloud in the relative coordinate system into the absolute coordinate system; Step 6: Based on the converted point cloud, capture the point cloud slices in the view, adjust the slices to the top view state, draw the floor plan of the underground pipeline corridor along the point cloud trajectory, overlay the floor plan with the topographic map, and obtain the real spatial location information of the underground pipeline corridor.

[0006] In the above-mentioned underground pipe gallery detection method based on drone point cloud scanning, in step 1, the ground reflector is a square reflector with a side length of 1 decimeter, and there are 4 ground reflectors, which are evenly distributed in a quadrilateral in space according to the size of the underground pipe gallery.

[0007] In the above-mentioned underground pipe gallery detection method based on drone point cloud scanning, in step 3, the relative coordinate system for establishing spatial feature association includes dynamically updating the relative position of the drone in space based on the Kalman filter algorithm to construct a three-dimensional map of the environment.

[0008] In the above-mentioned underground pipe gallery detection method based on drone point cloud scanning, in step 3: the state update is expressed as: ,in Indicates the current drone status. Indicates the drone status at the last positioning moment. The control input representing the current state, Indicates noise; The observation update is expressed as: ,in represents the observed value, Represents an environment map, represents the observation noise.

[0009] In the above-mentioned underground pipeline corridor detection method based on drone point cloud scanning, step 3 includes: by continuously iterating the state update equation and the observation update equation, the SLAM algorithm simultaneously optimizes the environment map With drone location , dynamically updates the relative position of the drone in space and builds a three-dimensional map of the environment.

[0010] In the above-mentioned underground pipe gallery detection method based on drone point cloud scanning, in step 5, the conversion parameters include three translation parameters: , scaling parameters , three rotation parameters , the conversion process includes translation, rotation, and scaling.

[0011] In the above-mentioned underground pipe gallery detection method based on drone point cloud scanning, step 5 comprises: Step 5-1: Build a transformation process model: ,in Represents the rotation matrix, which determines the parameters of the four ground reflectors that exist in both the absolute space coordinate system and the relative space coordinate system; Step 5-2: Construct the antisymmetric matrix S, ,in represents the independent elements of the Rodriguez matrix; Step 5-3: The rotation matrix R is constructed from the antisymmetric matrix S to form the Rodriguez matrix M: , , ; Step 5-4: Subtract the equations of two adjacent ground reflectors, eliminate the translation parameter, and obtain the property equation based on the properties of the antisymmetric matrix and the Rodriguez matrix: ,in is the third-order identity matrix, represents the matrix transpose, represents matrix inversion; Step 5-5: Substitute the property equation into the equation for subtracting two adjacent ground reflection pieces to obtain the subtraction equation: ; Step 5-6: Order , sorting out the subtraction equations to get the final equation: ; Step 5-7: Repeat steps 5-1 to 5-6 to calculate the final equation between the remaining two adjacent ground reflection pieces, combine the two final equations, solve the independent parameters of the transformation matrix, substitute the scale parameter and the rotation parameter into the transformation process model, and solve the translation parameter; Step 5-8: Apply the transformation process model point by point to complete the transformation of all point cloud data in the relative coordinate system obtained by the drone to the absolute coordinate system.

[0012] The beneficial effect of the underground pipe gallery detection method based on drone point cloud scanning of the present invention is as follows: the present invention uses a drone equipped with a laser radar to continuously scan the space from the ground to the underground pipe gallery, and combines ground GNSS measurement, and applies the Rodrigues matrix conversion method to obtain the real spatial position of the underground pipe gallery in the absolute coordinate system, providing reliable data support for urban planning and underground pipe gallery maintenance and transformation. The signal extension line is used to solve the communication problem between the drone and the ground remote control when flying in the underground pipe gallery; the drone continuously flies from the ground to the inside of the underground pipe gallery, and based on the SLAM algorithm, the point cloud data of the continuous space from the ground to the underground pipe gallery is calculated. By arranging 4 reflectors near the entrance of the ground pipe gallery, and using GNSS measurement to obtain its three-dimensional coordinates in the absolute coordinate system, the corresponding relative coordinates of the reflectors are extracted in the drone scanning point cloud, and the Rodrigues matrix conversion method is applied to match the relative point cloud to the absolute coordinate system, thereby obtaining the real spatial position information of the underground pipe gallery. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 A schematic diagram of on-site operations for the implementation of the present invention; Figure 2 The present invention is implemented in the field operation flow chart. DETAILED DESCRIPTION

[0014] In order to enable those skilled in the art to better understand the technical solution of the present invention, the technical solution of the present invention is described below in conjunction with specific implementation methods and drawings.

[0015] Example 1 In recent years, with the development of UAV technology, UAVs integrated with laser radar have gradually become an important tool for space exploration. In order to solve the problem that traditional measurement methods are difficult to comprehensively and accurately obtain the real spatial position of underground pipeline corridors in underground pipeline corridor detection, especially for underground pipeline corridors with great burial depth and harsh environment, the present invention proposes an underground pipeline corridor detection method that uses a UAV laser radar to scan and collect continuous spatial point cloud data above and below the ground, and combines it with ground GNSS high-precision positioning technology.

[0016] A method for detecting underground pipe corridors based on unmanned aerial vehicle point cloud scanning includes at least the following steps.

[0017] 1. Ground reflector layout and GNSS measurement. Four reflectors are laid out on the ground near the manhole cover of the underground corridor, and the reflectors are evenly distributed in a quadrilateral in space. The 3D coordinates of the four reflectors in the absolute coordinate system are collected using GNSS receivers using RTK mode and other operating methods.

[0018] 2. UAV flight operation and 3D laser scanning. Prepare the UAV for takeoff on the ground. Install the laser radar on the UAV body, connect the signal extension line to the remote control, start the UAV, first remotely control the UAV to fly on the ground, and simultaneously perform 3D laser scanning so that the scanning range covers 4 reflectors. Then remotely control the UAV to fly into the underground corridor through the manhole cover. At this time, place the signal extension line into the corridor, observe the internal situation of the corridor through the high-definition camera on the UAV body, and remotely control the UAV to fly along the corridor. After the flight mission is completed, the remotely controlled UAV returns to the ground to land and stops flying.

[0019] 3. Point cloud SLAM solution. During the flight of the drone, the lidar collects spatial point cloud data, while the visual sensor of the fuselage records environmental image information. The visual sensor extracts feature points in the environment, and combines the point cloud features of the lidar to establish spatial feature associations. Then, based on the Kalman filter algorithm, the relative position of the drone in space is dynamically updated, and a three-dimensional map of the environment is constructed.

[0020] 4. Extract the coordinates of the reflector from the point cloud scanned by the drone. After solving the continuous spatial point cloud above and below the ground based on the SLAM algorithm, the three-dimensional coordinates of the four reflectors are manually extracted using the difference in reflection intensity between the reflector and the surrounding environment in the point cloud. Because the point cloud is obtained in a relative coordinate system, the extracted three-dimensional coordinates of the reflector are also the results of the relative coordinate system.

[0021] 5. Use the Rodrigues matrix conversion method to convert the relative coordinate system point cloud scanned by the drone to the absolute coordinate system. Based on the three-dimensional coordinates of the reflector in the absolute coordinate system measured by the GNSS receiver on site and the three-dimensional coordinates of the reflector in the relative coordinate system extracted from the drone point cloud, use the Rodrigues matrix conversion method to convert the relative point cloud obtained by the drone to the absolute coordinate system.

[0022] 6. Drawing of plan views and other drawings. Based on the converted point cloud, cut a point cloud slice of appropriate thickness in the side view, adjust the slice to the top view state, and draw the plan view of the underground pipeline corridor along the point cloud trajectory. Use the same method to draw cross-section views and other drawings. Overlay the plan view with the topographic map to obtain the true spatial location information of the underground pipeline corridor.

[0023] Example 2 like Figure 1-Figure 2As shown, a method for underground pipeline corridor detection based on drone point cloud scanning includes the following steps.

[0024] Step 1: Deploy ground reflectors and use a GNSS receiver in PTK mode to collect the three-dimensional coordinates of the reflectors in an absolute coordinate system.

[0025] Four square reflectors with a side length of 1 decimeter are laid on the ground near the manhole cover of the underground corridor, such as Figure 1 The four reflectors A, B, C, and D are evenly distributed in a quadrilateral in space, and the spacing is moderate. If the spacing is too far, the subsequent drone flight scanning time will be long, affecting the effective operation flight time. If the spacing is too short, it will affect the subsequent coordinate conversion accuracy. Use a GNSS receiver to use RTK mode and other operating methods to collect the three-dimensional coordinates of the four reflectors in the absolute coordinate system: , , , .

[0026] Step 2: Prepare the drone for takeoff on the ground. Install the laser radar on the drone body, connect the signal extension line to the remote control, start the drone, first remotely control the drone to fly on the ground, and simultaneously perform 3D laser scanning to collect point clouds around the ground at the entrance of the underground corridor manhole cover, so that the scanning range covers 4 reflectors. Then remotely control the drone to fly into the underground corridor through the manhole cover. At this time, place the terminal of the signal extension line in the corridor, such as Figure 1 In the figure, the drone uses a high-definition camera on the fuselage to observe the internal situation of the tunnel, and remotely controls the drone to fly along the tunnel. After the flight mission is completed, the drone returns to the ground and stops flying. The laser radar carried by the drone collects spatial point cloud data, and the visual sensor records environmental image information.

[0027] Step 3: Point cloud SLAM solution. The feature points in the environment are extracted by visual sensors, and the point cloud features of the lidar are combined to establish a relative coordinate system associated with spatial features. Point cloud SLAM solution is performed based on state updates and observation updates. Establishing a relative coordinate system associated with spatial features includes dynamically updating the relative position of the drone in space based on the Kalman filter algorithm and building a three-dimensional map of the environment. During the flight of the drone, the laser radar collects spatial point cloud data, while the visual sensor of the fuselage records the environmental image information. The visual sensor extracts the feature points in the environment, and combines the point cloud features of the laser radar to establish spatial feature associations. In this process, the core of SLAM positioning is to solve the two problems of state update and observation update. The state update can be expressed as: (1), In formula (1), is the current drone status, is the drone status at the last positioning moment, is the control input of the current state, For noise.

[0028] The observation update can be expressed as: (2), In formula (2), is the observed value, For the environment map, is the observation noise. By continuously iterating the above two equations, the SLAM algorithm can simultaneously optimize the environment map With drone location , thereby dynamically updating the relative position of the drone in space and building a three-dimensional map of the environment.

[0029] Step 4: Extract the coordinates of the reflector from the scanned point cloud, calculate the continuous spatial point cloud above and below the ground based on the SLAM algorithm, and use the difference in reflection intensity between the reflector and the surrounding environment in the point cloud to extract the three-dimensional coordinates of the reflector in the relative coordinate system.

[0030] Manually extract the 3D coordinates of the 4 reflectors: , , , Since the point cloud is obtained in the relative coordinate system, the extracted three-dimensional coordinates of the reflector are also the results in the relative coordinate system.

[0031] Step 5: Determine the transformation parameters and transformation process, and apply the Rodriguez matrix transformation method to transform the point cloud in the relative coordinate system to the absolute coordinate system.

[0032] To convert the point cloud collected by the drone in the relative coordinate system to the absolute coordinate system, translation, rotation, scaling and other processes are required. , Seven parameters.

[0033] Specifically include: Step 5-1: Establish a conversion process model. The conversion process can be represented by a mathematical model as follows: (3), In formula (3), Represents a rotation matrix consisting of three rotation parameters constitute, Respectively represent the rotation angles of the three coordinate axes X, Y, and Z. Determine the parameters of the four ground reflectors that exist in the absolute space coordinate system and the relative space coordinate system. There are four common points A, B, C, and D in the two coordinate systems. The seven parameters are uniquely determined and can be directly solved. The nine elements in the rotation matrix R are determined by three angles, so only three are independent.

[0034] Step 5-2: Construct the antisymmetric matrix S, (4), where Represents the independent elements of the Rodriguez matrix.

[0035] Step 5-3: The rotation matrix R is constructed from the antisymmetric matrix S to form the Rodriguez matrix M: (5), In formula (5), (6), (7).

[0036] When calculating, first calculate the scale parameter, then solve the rotation matrix, and finally calculate the translation parameter. The scale parameter is calculated by taking the coordinates of two common points in the two coordinate systems, inversely calculating the ratio of the corresponding side lengths, and repeating the calculation for multiple pairs of common points and taking the average to improve the accuracy of the scale parameter. From formula (3), it can be seen that a set of three equations can be listed for each pair of common points, and 12 equations can be listed for the four reflectors. The equation of reflector B is subtracted from the equation of reflector A to eliminate the translation parameter, and the subsequent calculation is performed based on the properties of the antisymmetric matrix and the Rodriguez matrix.

[0037] Step 5-4: Subtract the equations of two adjacent ground reflectors, eliminate the translation parameter, and obtain the property equation based on the properties of the antisymmetric matrix and the Rodriguez matrix: ,in is the third-order identity matrix, represents the matrix transpose, It represents the inverse of the matrix, which is the inverse matrix of the original matrix.

[0038] Step 5-5: Substitute the property equation into the equation for subtracting two adjacent ground reflection pieces to obtain the subtraction equation: .

[0039] Step 5-6: Order , sorting out the subtraction equations to get the final equation: .

[0040] Then, by subtracting the equation listed on reflector C from the equation listed on reflector D, a set of equations similar to equation (11) can be obtained. By combining equation (11), the three independent parameters of the transformation matrix can be solved. Finally, the scale parameter and rotation parameter are substituted into equation (3) to solve the translation parameter. All point cloud data in the relative coordinate system obtained by the drone are transformed to the absolute coordinate system by applying the transformation formula (3) point by point.

[0041] Specific: Step 5-7: Repeat steps 5-1 to 5-6, calculate the final equation between the remaining two adjacent ground reflectors, combine the two final equations, solve the independent parameters of the transformation matrix, substitute the scale parameter and the rotation parameter into the transformation process model, and solve the translation parameter.

[0042] Step 5-8: Apply the transformation process model point by point to complete the transformation of all point cloud data in the relative coordinate system obtained by the drone to the absolute coordinate system.

[0043] Step 6: Drawing of floor plans and other drawings. Based on the converted point cloud, cut out the point cloud slices in the view, adjust the slices to the top view state, draw the floor plan of the underground pipeline corridor along the point cloud trajectory, overlay the floor plan with the topographic map, and obtain the real spatial location information of the underground pipeline corridor.

[0044] Specifically, a point cloud slice of appropriate thickness is captured in the side view, and after adjusting the slice to a top view state, the plan view of the underground pipeline corridor is drawn along the point cloud trajectory. The cross-sectional view and other drawings are drawn using the same method. The plan view is superimposed on the topographic map to obtain the true spatial location information of the underground pipeline corridor.

[0045] The above embodiments are only for illustrating the inventive concept and features of the present invention, and their purpose is to enable those skilled in the art to understand the content of the present invention and implement it accordingly, and they cannot be used to limit the protection scope of the present invention. Any equivalent changes or modifications made based on the essence of the content of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for detecting underground pipe corridors based on drone point cloud scanning, characterized in that: The following steps are involved: Step 1: Lay out ground reflectors and use a GNSS receiver in PTK mode to collect the three-dimensional coordinates of the reflectors in an absolute coordinate system; Step 2: Install a laser radar on the drone, connect the signal extension line to the remote control, and perform 3D laser scanning while the drone is flying to collect point cloud data at the entrance of the underground corridor. The scanning range covers all reflectors, and the terminal of the signal extension line is placed in the corridor. The drone is controlled to enter the underground corridor through the entrance. The laser radar carried by the drone collects spatial point cloud data, and the visual sensor records environmental image information. Step 3: Extract feature points in the environment through visual sensors, combine them with the point cloud features of the lidar, establish a relative coordinate system associated with spatial features, and perform point cloud SLAM solution based on state updates and observation updates; Step 4: Calculate the continuous spatial point cloud above and below the ground based on the SLAM algorithm, and use the difference in reflection intensity between the reflector and the surrounding environment in the point cloud to extract the three-dimensional coordinates of the reflector in the scanned point cloud in the relative coordinate system; Step 5: Determine the transformation parameters and transformation process, and apply the Rodrigues matrix transformation method to transform the point cloud in the relative coordinate system into the absolute coordinate system; Step 6: Based on the converted point cloud, capture the point cloud slices in the view, adjust the slices to the top view state, draw the floor plan of the underground pipeline corridor along the point cloud trajectory, overlay the floor plan with the topographic map, and obtain the real spatial location information of the underground pipeline corridor.

2. The underground pipe gallery detection method based on drone point cloud scanning according to claim 1 is characterized in that: In step 1, the ground reflector is a square reflector with a side length of 1 decimeter. There are 4 ground reflectors, and the 4 ground reflectors are evenly distributed in a quadrilateral in space according to the size of the underground pipe gallery.

3. The underground pipe gallery detection method based on drone point cloud scanning according to claim 1 is characterized in that: In step 3, the establishment of a relative coordinate system for associating spatial features includes dynamically updating the relative position of the drone in space based on a Kalman filter algorithm to construct a three-dimensional map of the environment.

4. The underground pipe gallery detection method based on drone point cloud scanning according to claim 3 is characterized in that: In step 3: the status update is expressed as: ,in Indicates the current drone status. Indicates the drone status at the last positioning moment. The control input representing the current state, Indicates noise; The observation update is expressed as: ,in represents the observed value, Represents an environment map, represents the observation noise.

5. The underground pipe gallery detection method based on drone point cloud scanning according to claim 4 is characterized in that: The step 3 includes: by continuously iterating the state update equation and the observation update equation, the SLAM algorithm simultaneously optimizes the environment map With drone location , dynamically updates the relative position of the drone in space and builds a three-dimensional map of the environment.

6. The underground pipe gallery detection method based on drone point cloud scanning according to claim 2 is characterized in that: In step 5, the conversion parameters include three translation parameters , scaling parameters , three rotation parameters , the conversion process includes translation, rotation, and scaling.

7. The underground pipe gallery detection method based on drone point cloud scanning according to claim 6 is characterized in that: The step 5 comprises: Step 5-1: Build a transformation process model: ,in Represents the rotation matrix, which determines the parameters of the four ground reflectors that exist in both the absolute space coordinate system and the relative space coordinate system; Step 5-2: Construct the antisymmetric matrix S, ,in represents the independent elements of the Rodriguez matrix; Step 5-3: The rotation matrix R is constructed from the antisymmetric matrix S to form the Rodriguez matrix M: ， ， ； Step 5-4: Subtract the equations of two adjacent ground reflectors, eliminate the translation parameter, and obtain the property equation based on the properties of the antisymmetric matrix and the Rodriguez matrix: ,in is the third-order identity matrix, represents the matrix transpose, represents matrix inversion; Step 5-5: Substitute the property equation into the equation for subtracting two adjacent ground reflection pieces to obtain the subtraction equation: ； Step 5-6: Order , sorting out the subtraction equations to get the final equation: ； Step 5-7: Repeat steps 5-1 to 5-6 to calculate the final equation between the remaining two adjacent ground reflection pieces, combine the two final equations, solve the independent parameters of the transformation matrix, substitute the scale parameter and the rotation parameter into the transformation process model, and solve the translation parameter; Step 5-8: Apply the transformation process model point by point to complete the transformation of all point cloud data in the relative coordinate system obtained by the drone to the absolute coordinate system.

Images