A pipeline splicing method based on laser radar vision assistance

Abstract

The present application relates to a pipeline splicing method based on laser radar visual assistance, comprising the following steps: S1, point cloud data acquisition: using a solid-state laser radar installed at the front end of a pipeline carrier to collect three-dimensional point cloud data of pipelines and tunnels in real time, and performing noise reduction processing on the collected three-dimensional point cloud data; S2, pipeline feature information extraction: based on the three-dimensional point cloud data of pipelines and tunnels after noise reduction, design pipeline recognition and feature extraction algorithms, complete the fitting of the central axis of the vehicle-mounted pipeline and the central axis of the laid pipeline; S3, pipeline posture adjustment: based on the extracted pipeline feature information, calculate the posture transformation matrix between the vehicle-mounted pipeline and the laid pipeline, and send a pipeline posture adjustment instruction to the control unit of the pipeline carrier to adjust the posture of the vehicle-mounted pipeline to meet the splicing accuracy requirements. The present application significantly enhances the accuracy and efficiency of pipeline splicing operations.

Description

Technical Field

[0001] The present invention relates to the field of pipeline splicing, and in particular to a pipeline splicing method based on laser radar vision assistance. Background Art

[0002] At present, during pipeline laying, especially the installation and construction of pipelines in tunnels, multiple pipe sections need to be spliced. Traditional tunnel pipeline splicing operations rely on manual measurement and adjustment, and have problems such as low automation, low efficiency and insufficient precision. Summary of the invention

[0003] In view of the defects of the prior art, the present invention provides a pipeline splicing method based on laser radar vision assistance.

[0004] In order to achieve the above object, the technical solution adopted by the present invention is: a pipeline splicing method based on laser radar vision assistance, characterized in that it includes the following steps:

[0005] S1. Point cloud data acquisition: Use the solid-state laser radar installed at the front end of the pipeline carrier to collect the three-dimensional point cloud data of the pipeline and tunnel in real time, and perform noise reduction on the collected three-dimensional point cloud data;

[0006] S2. Extraction of pipeline feature information: Based on the 3D point cloud data of pipelines and tunnels after noise reduction, pipeline recognition and feature extraction algorithms are designed to complete the fitting of the center axis of the vehicle-mounted pipeline and the center axis of the laid pipeline;

[0007] S3. Pipeline posture adjustment: Based on the extracted pipeline feature information, the posture transformation matrix between the vehicle-mounted pipeline and the laid pipeline is calculated, and the pipeline posture adjustment instruction is sent to the control unit of the pipeline carrier to adjust the posture of the vehicle-mounted pipeline to achieve the coincidence of the central axis of the vehicle-mounted pipeline with the central axis of the laid pipeline to meet the splicing accuracy requirements.

[0008] As a preferred solution, step S1 includes the following sub-steps:

[0009] S11. Data acquisition: The solid-state laser radar is installed at the front end of the pipeline carrier. Its shooting angle is along the forward direction of the pipeline carrier and tilted upward. The radar system emits laser pulses at a high frequency and receives laser signals reflected from the surrounding environment. By measuring the round-trip time of the light pulse, the radar can accurately calculate the distance to each point irradiated by the laser, thereby constructing the three-dimensional point cloud data of the pipeline and tunnel in real time;

[0010] S12, through-filter denoising: For each frame of point cloud data collected by the solid-state laser radar, a through-filter algorithm is used to remove point cloud data that does not belong to pipelines and tunnels, and retain point cloud data within the limited range of each dimension. The calculation formula is as follows:

[0011]

[0012] in, , , They are x , y , z Direction thresholds;

[0013] S13. Outlier removal: Use statistical filtering algorithm to remove outliers in point cloud data.

[0014] As a preferred solution, step S13 includes the following steps:

[0015] Calculate the average distance between each point in the point cloud data and its surrounding neighborhood points ;

[0016] Construct a Gaussian distribution based on the average distance of all points. The mean of the Gaussian distribution is and standard deviation The calculation formula is as follows:

[0017]

[0018] Among them, n is the total number of points in the neighborhood point set;

[0019] All average distances greater than The points are regarded as outliers and removed from the original point cloud data, where is the set positive threshold coefficient.

[0020] As a preferred solution, step S2 specifically includes the following steps:

[0021] S21, point cloud segmentation: The collected pipeline and tunnel point clouds include vehicle-mounted pipeline point clouds, tunnel point clouds and laid pipeline point clouds. The collected pipeline and tunnel point clouds are segmented, which specifically includes the following steps:

[0022] S211, calculating the Euclidean distance between each point in the pipeline and tunnel point cloud, and using the Euclidean clustering algorithm to divide the original pipeline and tunnel point cloud into multiple point cloud sets, each point cloud set must meet the set minimum point number requirement;

[0023] S212, when the pipeline carrier has not yet reached the pipeline splicing work area, the solid-state laser radar will collect the on-board pipeline point cloud and the tunnel point cloud, generate two cluster point clouds through the Euclidean clustering algorithm, calculate the centroids of the two cluster point clouds respectively, and designate the point clouds closest and second closest to the solid-state laser radar as the on-board pipeline point cloud and the tunnel point cloud respectively;

[0024] S213, when the pipeline carrier arrives at the working area for pipeline splicing, the solid-state laser radar collects the on-board pipeline point cloud, the tunnel point cloud, and the laid pipeline point cloud, generates three cluster point clouds through the Euclidean clustering algorithm, designates the point cloud closest to the solid-state laser radar as the on-board pipeline point cloud, and designates the point clouds closest and second closest to the centroid of the on-board pipeline point cloud as the tunnel point cloud and the laid pipeline point cloud, respectively;

[0025] S22, rough extraction of the pipeline centerline: The centerline of the vehicle-mounted pipeline and the centerline of the laid pipeline are roughly extracted by the RANSAC cylindrical fitting algorithm, which specifically includes the following steps: using the RANSAC cylindrical fitting algorithm to perform cylindrical surface fitting on the pipeline point cloud, the pipeline point cloud is the vehicle-mounted pipeline point cloud or the laid pipeline point cloud obtained in step S213, and using the RANSAC cylindrical fitting algorithm to perform cylindrical surface fitting on the pipeline point cloud, and the cylindrical surface formula is expressed as:

[0026]

[0027] in, represents a point on the cylindrical surface, represents a point on the axis of the cylinder, represents the direction vector of the cylinder axis, Represents the radius of the cylinder. The RANSAC cylinder fitting algorithm fits the cylindrical surface specifically including the following steps:

[0028] S221. Randomly select three spatial points from the pipeline point cloud each time to determine a cylindrical surface model, and solve the central axis direction and radius of the cylindrical surface;

[0029] S222, calculating the distance from the remaining point set to the cylindrical surface, and solving the inner point set belonging to the cylindrical surface model according to a preset distance threshold;

[0030] S223, iterating the above process multiple times, continuously optimizing the selected internal point set and cylindrical surface model, and finally determining the fitting cylinder with the largest number of internal points as the optimal fitting model, thereby completing the fitting of the cylindrical surface of the pipeline point cloud and the rough extraction of its central axis;

[0031] S23, accurate extraction of the central axis of the pipeline: Correcting the central axis of the vehicle-mounted pipeline extracted in step S22, specifically including the following steps:

[0032] S231, along the central axis of the vehicle-mounted pipeline extracted in step S22 Direction, Equal Interval Slice the vehicle pipeline point cloud, and the slicing plane is perpendicular to the vehicle centerline. For any slicing plane, calculate the distance between the vehicle pipeline point cloud and the slicing plane, and designate the point set less than the set distance threshold as the point in the slicing plane, and project the points in the slicing plane onto the slicing plane;

[0033] S232. For each projected point cloud on the tangent plane, the two-dimensional projected discrete point set is fitted into an elliptic curve using the least squares method. The fitting equation is:

[0034]

[0035] in, , denote the major and minor semi-axes of the ellipse, respectively. Represents the coordinates of the center of the ellipse;

[0036] S233, solving the centroid of each section ellipse calculated in step S232 and The average value of the direction is expressed as ;

[0037] S234, the central axis of the vehicle-mounted pipeline extracted in step S22 is offset along its vertical direction by a vector Move to accurately extract the centerline of the vehicle-mounted pipeline;

[0038] Then, the central axis of the laid pipeline extracted in step S22 is corrected by the same steps as steps S231 to S234 above, so as to complete the accurate extraction of the central axis of the laid pipeline;

[0039] S24. Extraction of the position of the pipeline end face: The vehicle-mounted pipeline point cloud obtained in step S213 includes the end face point cloud information of the vehicle-mounted pipeline, and the laid pipeline point cloud includes the end face point cloud information of the laid pipeline. According to the end face point cloud information of the vehicle-mounted pipeline and the end face point cloud information of the laid pipeline, the positions of the vehicle-mounted pipeline end face and the laid pipeline end face are extracted, specifically including the following steps: for the vehicle-mounted pipeline or the laid pipeline, its pipeline end face point cloud is located at the boundary of the collected point cloud, and the boundary point cloud is extracted by the point cloud boundary extraction algorithm. Based on the boundary point cloud extracted in step S1, the RANSAC plane fitting algorithm is used to complete the fitting of the pipeline end face, and the spatial position parameters of the pipeline end face are solved.

[0040] As a preferred solution, step S3 specifically includes the following steps:

[0041] S31, calculation of rotation matrix: according to the central axis posture of the vehicle-mounted pipeline, the spatial position parameters of the end face of the vehicle-mounted pipeline, the central axis posture of the laid pipeline, and the spatial position parameters of the end face of the laid pipeline extracted in step S2, the posture transformation matrix between the vehicle-mounted pipeline and the laid pipeline is calculated. The posture transformation matrix represents , including the rotation matrix and translation matrix Calculation; Assume that at a certain moment, for the vehicle-mounted pipeline and the laid pipeline, the center axis of the vehicle-mounted pipeline and the center axis of the laid pipeline represent and ,in, and express and The direction vector of and The angle between , the calculation formula is as follows:

[0042]

[0043] based on and The angle between Constructing the rotation matrix , the calculation formula is as follows:

[0044]

[0045] in, express The rotation axis is calculated as , using the rotation matrix The center axis of the vehicle pipeline Transform to the centerline of the laid pipe parallel;

[0046] S32, translation matrix calculation: calculate the centerline of the vehicle-mounted pipeline after rotation transformation Any point on the pipeline to the centerline of the laid pipeline The spatial vector pointing to the projection point is the required translation vector , according to the required rotation matrix and translation matrix , the transformation matrix from the vehicle-mounted pipeline to the laid pipeline can be obtained , the calculation formula is ;

[0047] S33, pipeline posture adjustment: the pipeline posture transformation matrix obtained in step S32 is The parameters are sent to the pipe posture adjustment device of the pipeline carrier to adjust the spatial position and posture of the vehicle-mounted pipeline to ensure that the central axis of the vehicle-mounted pipeline coincides with the central axis of the laid pipeline.

[0048] The beneficial effects of the present application are as follows: the present application performs pipeline splicing operations based on lidar vision, which improves the automation and intelligence level of pipeline laying, significantly enhances the accuracy and efficiency of pipeline splicing operations, and effectively overcomes the limitations of traditional pipeline splicing systems that rely on manual measurement and adjustment. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a schematic diagram of the structure of the pipeline carrier and solid-state laser radar for this application;

[0050] Figure 2 A schematic diagram of solid-state laser radar data collection when the pipeline carrier vehicle of this application has not yet reached the pipeline splicing work area;

[0051] Figure 3 This is a schematic diagram of solid-state laser radar data collection when the pipeline carrier reaches the pipeline splicing work area for this application.

[0052] Figure numerals: 1. transport trolley, 2. load-bearing beam, 3. pipeline posture adjustment device, 4. power unit, 5. solid-state laser radar, 6. vehicle-mounted pipeline, 7. tunnel, 8. laid pipeline. DETAILED DESCRIPTION

[0053] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0054] The present application embodiment provides a pipeline splicing method based on laser radar vision assistance, which includes the following steps:

[0055] S1. Point cloud data acquisition: using the solid-state laser radar 5 installed at the front end of the pipeline carrier to collect the three-dimensional point cloud data of the pipeline and the tunnel 7 in real time, and performing noise reduction processing on the collected three-dimensional point cloud data;

[0056] S2. Extraction of pipeline feature information: Based on the 3D point cloud data of the pipeline and tunnel 7 after noise reduction, a pipeline recognition and feature extraction algorithm is designed to complete the fitting of the center axis of the vehicle-mounted pipeline and the center axis of the laid pipeline;

[0057] S3, pipeline posture adjustment: Based on the extracted pipeline feature information, the posture transformation matrix between the vehicle-mounted pipeline 6 and the laid pipeline 8 is calculated, and a pipeline posture adjustment instruction is sent to the control unit of the pipeline carrier to adjust the posture of the vehicle-mounted pipeline 6 to achieve the coincidence of the central axis of the vehicle-mounted pipeline with the central axis of the laid pipeline to meet the splicing accuracy requirements.

[0058] Step S1 includes the following sub-steps:

[0059] S11. Data acquisition: The solid-state laser radar 5 is installed at the front end of the load-bearing beam 2 of the pipeline carrier. Its shooting angle is along the forward direction of the pipeline carrier and tilted upward. The radar system emits laser pulses at a high frequency and receives laser signals reflected from the surrounding environment. By measuring the round-trip time of the light pulse, the radar can accurately calculate the distance to each point irradiated by the laser, thereby constructing the three-dimensional point cloud data of the pipeline and tunnel 7 in real time.

[0060] S12, through-filter noise reduction: For each frame of point cloud data collected by the solid-state laser radar 5, a through-filter algorithm is used to remove point cloud data that does not belong to the pipeline and tunnel 7, and retain point cloud data within the limited range of each dimension. The calculation formula is as follows:

[0061]

[0062] in, , , They are x , y , z Direction limit threshold.

[0063] S13. Outlier removal: Use statistical filtering algorithm to remove outliers in point cloud data.

[0064] Step S13 includes the following steps: calculating the average distance between each point in the point cloud data and its surrounding neighborhood point set ; Construct a Gaussian distribution based on the average distance of all points, the mean of the Gaussian distribution and standard deviation The calculation formula is as follows:

[0065]

[0066] Where n is the total number of points in the neighborhood point set; all points with average distance greater than The points are regarded as outliers and removed from the original point cloud data, where is the set positive threshold coefficient.

[0067] Step S2 specifically includes the following steps:

[0068] S21, point cloud segmentation: The collected pipeline and tunnel 7 point clouds include vehicle-mounted pipeline point clouds, tunnel point clouds and laid pipeline point clouds. The collected pipeline and tunnel 7 point clouds are segmented, specifically including the following steps:

[0069] S211. Calculate the Euclidean distance between each point in the pipeline and tunnel 7 point cloud, and use the Euclidean clustering algorithm to divide the original pipeline and tunnel 7 point cloud into multiple point cloud sets, each point cloud set must meet the set minimum point number requirement.

[0070] S212. When the pipeline carrier has not yet reached the pipeline splicing work area, the solid-state laser radar 5 will collect the vehicle-mounted pipeline point cloud and the tunnel point cloud, generate two cluster point clouds through the Euclidean clustering algorithm, calculate the centroids of the two cluster point clouds respectively, and designate the point clouds closest and second closest to the solid-state laser radar 5 as the vehicle-mounted pipeline point cloud and tunnel point cloud respectively.

[0071] S213. When the pipeline carrier arrives at the pipeline splicing work area, the solid-state laser radar 5 collects the vehicle-mounted pipeline point cloud, the tunnel point cloud, and the laid pipeline point cloud, and generates three cluster point clouds through the Euclidean clustering algorithm. The point cloud closest to the solid-state laser radar 5 is designated as the vehicle-mounted pipeline point cloud, and the point clouds closest and second closest to the centroid of the vehicle-mounted pipeline point cloud are designated as the tunnel point cloud and the laid pipeline point cloud, respectively.

[0072] S22, rough extraction of the pipeline centerline: The centerline of the vehicle-mounted pipeline and the centerline of the laid pipeline are roughly extracted by the RANSAC cylindrical fitting algorithm, which specifically includes the following steps: using the RANSAC cylindrical fitting algorithm to perform cylindrical surface fitting on the pipeline point cloud, the pipeline point cloud is the vehicle-mounted pipeline point cloud or the laid pipeline point cloud obtained in step S213, and the cylindrical surface formula is expressed as:

[0073]

[0074] in, represents a point on the cylindrical surface, represents a point on the axis of the cylinder, represents the direction vector of the cylinder axis, Represents the radius of the cylinder. The RANSAC cylinder fitting algorithm fits the cylindrical surface specifically including the following steps:

[0075] S221. Randomly select three spatial points from the original point cloud each time to determine a cylindrical surface model, and solve the central axis direction and radius of the cylindrical surface;

[0076] S222, calculating the distance from the remaining point set to the cylindrical surface, and solving the inner point set belonging to the cylindrical surface model according to a preset distance threshold;

[0077] S223, iterate the above process multiple times, continuously optimize the selected internal point set and cylindrical surface model, and finally determine that the fitting cylinder with the largest number of internal points is the optimal fitting model, so as to complete the fitting of the cylindrical surface of the pipeline point cloud and the rough extraction of its central axis.

[0078] S23, accurate extraction of the central axis of the pipeline: Since the pipeline is usually made of rolled and welded steel plates and is generally a thin-walled structure, it is affected by gravity during transportation and assembly, and the cross section of the pipeline may be elliptical. It is necessary to correct the central axis of the vehicle-mounted pipeline extracted in step S22, which specifically includes the following steps:

[0079] S231, along the central axis of the vehicle-mounted pipeline extracted in step S22 Direction, Equal Interval Slice the vehicle-mounted pipeline point cloud, and the slicing plane is perpendicular to the central axis. For any slicing plane, calculate the distance between the vehicle-mounted pipeline point cloud and the slicing plane, and designate the point set less than the set distance threshold as the point in the slicing plane, and project the point in the slicing plane onto the slicing plane;

[0080] S232. For each projected point cloud on the tangent plane, the two-dimensional projected discrete point set is fitted into an elliptic curve using the least squares method. The fitting equation is:

[0081]

[0082] in, , denote the major and minor semi-axes of the ellipse, respectively. Represents the coordinates of the center of the ellipse;

[0083] S233, solving the centroid of each section ellipse calculated in step S232 and The average value of the direction is expressed as ;

[0084] S234, the central axis of the vehicle-mounted pipeline extracted in step S22 is offset along its vertical direction by a vector The vehicle-mounted pipeline is moved to accurately extract the center axis of the vehicle-mounted pipeline. The accuracy of the extraction of the center axis of the vehicle-mounted pipeline is improved through correction.

[0085] Then, the central axis of the laid pipeline extracted in step S22 is corrected using the same steps as steps S231-S234 above to complete the accurate extraction of the central axis of the laid pipeline. During the correction, the central axis of the vehicle-mounted pipeline involved in steps S231-S234 is the central axis of the laid pipeline, and the point cloud of the vehicle-mounted pipeline is the point cloud of the laid pipeline.

[0086] S24. Extraction of the position of the pipeline end face: The vehicle-mounted pipeline point cloud obtained in step S213 includes the end face point cloud information of the vehicle-mounted pipeline 6, and the laid pipeline point cloud includes the end face point cloud information of the laid pipeline 8. According to the end face point cloud information of the vehicle-mounted pipeline 6 and the end face point cloud information of the laid pipeline 8, the positions of the vehicle-mounted pipeline end face and the laid pipeline end face are extracted, specifically including the following steps: For the vehicle-mounted pipeline 6 or the laid pipeline 8, its pipeline end face point cloud is located at the boundary of the collected point cloud, and the boundary point cloud is extracted by the point cloud boundary extraction algorithm. Based on the boundary point cloud extracted in step S1, the RANSAC plane fitting algorithm is used to complete the fitting of the pipeline end face, and the spatial position parameters of the pipeline end face are solved.

[0087] S3 specifically includes the following steps:

[0088] S31, calculation of rotation matrix: according to the central axis posture of the vehicle-mounted pipeline, the spatial position parameters of the end face of the vehicle-mounted pipeline, the central axis posture of the laid pipeline, and the spatial position parameters of the end face of the laid pipeline extracted in step S2, the posture transformation matrix between the vehicle-mounted pipeline 6 and the laid pipeline 8 is calculated. The posture transformation matrix represents , including the rotation matrix and translation matrix Calculation; Assume that at a certain moment, for the vehicle-mounted pipeline 6 and the laid pipeline 8, the center axis of the vehicle-mounted pipeline and the center axis of the laid pipeline are respectively represented by and ,in, and express and The direction vector of and The angle between , the calculation formula is as follows:

[0089]

[0090] based on and The angle between Constructing the rotation matrix , the calculation formula is as follows:

[0091]

[0092] in, express The rotation axis is calculated as , using the rotation matrix The center axis of the vehicle pipeline Transform to the centerline of the laid pipe parallel.

[0093] S32, translation matrix calculation: calculate the centerline of the vehicle-mounted pipeline after rotation transformation Any point on the pipeline to the centerline of the laid pipeline The spatial vector pointing to the projection point is the required translation vector , according to the required rotation matrix and translation matrix , the transformation matrix from the vehicle-mounted pipeline 6 to the laid pipeline 8 can be obtained , the calculation formula is .

[0094] S33, pipeline posture adjustment: the posture transformation matrix parameters obtained in step S32 are The pipe posture adjustment device 3 sent to the pipeline carrier is used to adjust the spatial position and posture of the vehicle-mounted pipeline 6 to ensure that the central axis of the vehicle-mounted pipeline coincides with the central axis of the laid pipeline.

[0095] The pipeline transport vehicle of the present application includes a load-bearing beam 2, a power unit 4 installed on the load-bearing beam 2, and a transport trolley 1 for supporting the load-bearing beam 2 to move. A pipeline posture adjustment device 3 is installed on the load-bearing beam 2. The pipeline transport vehicle of the present application adopts the working vehicle in the existing patent CN 111677534 A, and a solid-state laser radar 5 is installed on the load-bearing beam 2.

[0096] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technician familiar with this profession can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention without departing from the content of the technical solution of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A pipeline splicing method based on laser radar vision assistance, characterized in that: The steps include: S1. Point cloud data acquisition: Use the solid-state laser radar installed at the front end of the pipeline carrier to collect the three-dimensional point cloud data of the pipeline and tunnel in real time, and perform noise reduction on the collected three-dimensional point cloud data; S2. Extraction of pipeline feature information: Based on the 3D point cloud data of pipelines and tunnels after noise reduction, pipeline recognition and feature extraction algorithms are designed to complete the fitting of the center axis of the vehicle-mounted pipeline and the center axis of the laid pipeline; S3, Pipeline posture adjustment: Based on the extracted pipeline feature information, the posture transformation matrix of the vehicle-mounted pipeline and the laid pipeline is calculated, and the pipeline posture adjustment instruction is sent to the control unit of the pipeline carrier to adjust the posture of the vehicle-mounted pipeline to achieve the coincidence of the central axis of the vehicle-mounted pipeline with the central axis of the laid pipeline, so that it meets the splicing accuracy requirements; Step S1 includes the following sub-steps: S11. Data acquisition: The solid-state laser radar is installed at the front end of the pipeline carrier. Its shooting angle is along the forward direction of the pipeline carrier and tilted upward. The radar system emits laser pulses at a high frequency and receives laser signals reflected by the surrounding environment. By measuring the round-trip time of the light pulse, the radar can accurately calculate the distance to each point irradiated by the laser, thereby constructing the three-dimensional point cloud data of the pipeline and tunnel in real time; S12, through-filter denoising: For each frame of point cloud data collected by the solid-state laser radar, a through-filter algorithm is used to remove point cloud data that does not belong to pipelines and tunnels, and retain point cloud data within the limited range of each dimension. The calculation formula is as follows: in, , , They are x , y , z Direction thresholds; S13, outlier removal: use statistical filtering algorithm to remove outliers in point cloud data; Step S13 includes the following steps: Calculate the average distance between each point in the point cloud data and its surrounding neighborhood points ; Construct a Gaussian distribution based on the average distance of all points. The mean of the Gaussian distribution is and standard deviation The calculation formula is as follows: Among them, n is the total number of points in the neighborhood point set; All average distances greater than The points are considered as outliers and removed from the point cloud data, where is the set positive threshold coefficient; Step S2 specifically includes the following steps: S21. Point cloud segmentation: The pipeline and tunnel point clouds after noise reduction include vehicle-mounted pipeline point clouds, tunnel point clouds, and laid pipeline point clouds. Point cloud segmentation is performed on the pipeline and tunnel point clouds after noise reduction, specifically including the following steps: S211, calculating the Euclidean distance between each point in the pipeline and tunnel point cloud, and using the Euclidean clustering algorithm to divide the pipeline and tunnel point cloud into multiple point cloud sets, each point cloud set must meet the set minimum point number requirement; S212, when the pipeline carrier has not yet reached the pipeline splicing work area, the solid-state laser radar will collect the on-board pipeline point cloud and the tunnel point cloud, generate two cluster point clouds through the Euclidean clustering algorithm, calculate the centroids of the two cluster point clouds respectively, and designate the point clouds closest and second closest to the solid-state laser radar as the on-board pipeline point cloud and the tunnel point cloud respectively; S213, when the pipeline carrier arrives at the working area for pipeline splicing, the solid-state laser radar collects the on-board pipeline point cloud, the tunnel point cloud, and the laid pipeline point cloud, generates three cluster point clouds through the Euclidean clustering algorithm, designates the point cloud closest to the solid-state laser radar as the on-board pipeline point cloud, and designates the point clouds closest and second closest to the centroid of the on-board pipeline point cloud as the tunnel point cloud and the laid pipeline point cloud, respectively; S22, rough extraction of the pipeline centerline: The centerline of the vehicle-mounted pipeline and the centerline of the laid pipeline are roughly extracted by the RANSAC cylindrical fitting algorithm, which specifically includes the following steps: using the RANSAC cylindrical fitting algorithm to perform cylindrical surface fitting on the pipeline point cloud, the pipeline point cloud is the vehicle-mounted pipeline point cloud or the laid pipeline point cloud obtained in step S213, and the cylindrical surface formula is expressed as: in, represents a point on the cylindrical surface, represents a point on the axis of the cylinder, represents the direction vector of the cylinder axis, Represents the radius of the cylinder. The RANSAC cylinder fitting algorithm fits the cylindrical surface specifically including the following steps: S221. Randomly select three spatial points from the pipeline point cloud each time to determine a cylindrical surface model, and solve the central axis direction and radius of the cylindrical surface; S222, calculating the distance from the remaining point set to the cylindrical surface, and solving the inner point set belonging to the cylindrical surface model according to a preset distance threshold; S223, iterating the above process multiple times, continuously optimizing the selected internal point set and cylindrical surface model, and finally determining the fitting cylinder with the largest number of internal points as the optimal fitting model, thereby completing the fitting of the cylindrical surface of the pipeline point cloud and the rough extraction of its central axis; S23, accurate extraction of the pipeline centerline: Correcting the vehicle-mounted pipeline centerline extracted in step S22, specifically including the following steps: S231, along the central axis of the vehicle-mounted pipeline extracted in step S22 Direction, Equal Interval Slice the vehicle-mounted pipeline point cloud, and the slicing plane is perpendicular to the central axis of the vehicle-mounted pipeline. For any slicing plane, calculate the distance between the vehicle-mounted pipeline point cloud and the slicing plane, and designate the point set less than the set distance threshold as the point in the slicing plane, and project the point in the slicing plane onto the slicing plane; S232. For each projected point cloud on the tangent plane, the two-dimensional projected discrete point set is fitted into an elliptic curve using the least squares method. The fitting equation is: in, , denote the major and minor semi-axes of the ellipse, respectively. Represents the coordinates of the center of the ellipse; S233, solving the centroid of each section ellipse calculated in step S232 and The average value of the direction is expressed as ; S234, the central axis of the vehicle-mounted pipeline extracted in step S22 is offset along its vertical direction by a vector Move to accurately extract the centerline of the vehicle-mounted pipeline; Then, the central axis of the laid pipeline extracted in step S22 is corrected by the same steps as steps S231 to S234 above, so as to complete the accurate extraction of the central axis of the laid pipeline; S24, position extraction of pipeline end face: The vehicle-mounted pipeline point cloud obtained in step S213 includes the end face point cloud information of the vehicle-mounted pipeline, and the laid pipeline point cloud includes the end face point cloud information of the laid pipeline. According to the end face point cloud information of the vehicle-mounted pipeline and the end face point cloud information of the laid pipeline, the positions of the vehicle-mounted pipeline end face and the laid pipeline end face are extracted, specifically including the following steps: for the vehicle-mounted pipeline or the laid pipeline, its pipeline end face point cloud is located at the boundary of the collected point cloud, the boundary point cloud is extracted by using the point cloud boundary extraction algorithm, based on the extracted boundary point cloud, the pipeline end face is fitted using the RANSAC plane fitting algorithm, and the spatial position parameters of the pipeline end face are solved.

2. The pipeline splicing method based on laser radar vision assistance according to claim 1 is characterized in that: Step S3 specifically includes the following steps: S31, calculation of rotation matrix: according to the central axis posture of the vehicle-mounted pipeline, the spatial position parameters of the end face of the vehicle-mounted pipeline, the central axis posture of the laid pipeline, and the spatial position parameters of the end face of the laid pipeline extracted in step S2, the posture transformation matrix between the vehicle-mounted pipeline and the laid pipeline is calculated. The posture transformation matrix represents , including the rotation matrix and translation matrix Calculation; Assume that at a certain moment, for the vehicle-mounted pipeline and the laid pipeline, the center axis of the vehicle-mounted pipeline and the center axis of the laid pipeline represent and ,in, and express and The direction vector of and The angle between , the calculation formula is as follows: based on and The angle between Constructing the rotation matrix , the calculation formula is as follows: in, express The rotation axis is calculated as , using the rotation matrix The center axis of the vehicle pipeline Transform to the centerline of the laid pipe parallel; S32, translation matrix calculation: calculate the centerline of the vehicle-mounted pipeline after rotation transformation Any point on the pipeline to the centerline of the laid pipeline The spatial vector pointing to the projection point is the required translation vector , according to the required rotation matrix and translation matrix , the pose transformation matrix from the vehicle-mounted pipeline to the laid pipeline can be obtained , the calculation formula is ; S33, pipeline posture adjustment: the posture transformation matrix obtained in step S32 The parameters are sent to the pipe posture adjustment device of the pipeline carrier to adjust the spatial position and posture of the vehicle-mounted pipeline to ensure that the center axis of the vehicle-mounted pipeline coincides with the center axis of the laid pipeline.

Images