Inhaul cable damper connection abnormity robot detection task point automatic planning method

Abstract

The invention relates to an inhaul cable damper connection abnormity robot detection task point automatic planning method, which is characterized by comprising point cloud clustering and pose calculation, and comprises the following steps: S1, bridge deck maintenance road scene spot cloud acquisition; S2, specific height inhaul cable slice acquisition; s3, clustering the extension starting points of the inhaul cables; s4, inhaul cable and damper clustering segmentation based on a density point cloud algorithm; s5, inhaul cable point cloud stripping; s6, task point coordinates and damper top end coordinates are obtained; and S7, the robot collection angle is calculated. Compared with the prior art, the method has the advantages of high efficiency, accuracy, adaptability to complex scenes and the like, and has the advantages of automatic task point planning, rapid data processing, accurate point cloud analysis and accurate task point positioning. Not only are the inspection efficiency and accuracy improved, but also the non-standardization degree of manual inspection is reduced, and powerful technical support is provided for safe operation of the bridge.

Description

Technical Field

[0001] The present invention relates to the field of automatic detection technology, and in particular to an automatic planning method for robot detection task points of cable damper connection anomalies. Background Art

[0002] As an important transportation infrastructure, the safe operation of bridges is of vital importance. The cable damper located on the bridge deck inspection road is a key component of the cable-stayed bridge, and its working condition directly affects the stability and durability of the bridge. However, traditional manual inspection methods have problems such as low standardization and insufficient frequency, which makes it difficult to meet the needs of safe bridge operation. In addition, the poor reproducibility of manual inspections makes it difficult to ensure the consistency and accuracy of data collection. Summary of the invention

[0003] The purpose of the present invention is to overcome the defects of the above-mentioned prior art such as low standardization level and insufficient frequency, and to provide an automatic planning method for robot detection task points of cable damper connection abnormalities, which not only improves the efficiency and accuracy of inspections, but also reduces the non-standardization degree of manual inspections, providing strong technical support for the safe operation of bridges.

[0004] The present invention can automatically plan the robot's task target point and the optimal angle for collecting the top connection status of the damper according to the acquired three-dimensional point cloud file of the bridge deck inspection road scene, thereby realizing automated and standardized data collection.

[0005] The present invention provides a method for automatically planning a cable damper connection abnormality robot detection task point, including point cloud clustering and posture calculation, and comprising the following steps:

[0006] S1: Obtaining point cloud of bridge deck inspection road scene: Through the FAST_LIO 3D SLAM algorithm embedded in the hardware device, the operator independently controls the hardware device remotely, so that the hardware device moves along the entire length of the bridge deck inspection road. After completion, the pcd file is saved to obtain the 3D point cloud scene file of the bridge deck inspection road. The hardware device can be independently operated and used by the operator. When the bridge deck inspection road point cloud scene is first established, the operator is required to manually remotely collect point clouds. Subsequent inspections can be automated based on existing maps.

[0007] S2: Obtaining cable slices at a specific height: Based on the normal estimation method of random sampling consensus (RANSAC), filter the data, estimate the normal, separate the model, and select the slices at the height above the street lamp to retain the cable point cloud;

[0008] S3: Clustering of cable extension starting points: The point cloud clusters near the same height of each cable are clustered by the k_means clustering method, and an actual point closest to its geometric center is used as the cable extension starting point;

[0009] S4: Cluster segmentation of cables and dampers based on density point cloud algorithm: Starting from the starting point of cable extension, based on the density clustering algorithm, the point cloud is expanded to include cables and dampers;

[0010] S5: Cable point cloud stripping: The spatial linear equation of each cable is obtained by analyzing the cable slice clustering results at different heights in S3. All point clouds with the spatial linear equation of each cable as the reference and the cable radius as the threshold are deleted from the point cloud obtained in S4, and then a point cloud file with only the damper is obtained;

[0011] S6: Obtaining the coordinates of the task point and the top coordinates of the damper: The coordinates of the anchor end are obtained from the linear equation of the cable space, and the coordinates of the task point a are obtained by combining the reserved space; the coordinates of the top of the damper b are obtained by clustering the damper point cloud file;

[0012] S7: Calculation of robot acquisition angle: The unique spatial straight line equation is determined by a and b, and the angle between it and the three axes in the point cloud file coordinate system is obtained to obtain the robot's horizontal rotation angle and pitch angle here, thereby realizing automatic planning of detection task points.

[0013] Furthermore, in S1, the hardware device includes: a perception module, a motion module, a control module and a communication module.

[0014] Furthermore, the perception module is a 3D laser radar with a built-in IMU, which is located at the top of the device; the control module includes a control main board and a switch, which are located in the upper shell, and an optical camera is also provided in the shell; the communication module is a built-in traffic card router, which can realize data transmission; the motion module is located at the bottom and is a wheeled chassis.

[0015] Furthermore, in S2, the normal estimation method based on random sampling consensus (RANSAC) is specifically as follows: filtering out data far away from the area of ​​interest; estimating the surface normal of each point; separating the plane model and the cylindrical model from the point cloud using the normal estimation method; and retaining and saving all the cable point clouds by selecting the height above the street lamp for slicing.

[0016] Further, in S3, cluster convergence judgment: repeat the above steps of distance calculation, cluster allocation and cluster center update until the cluster center no longer changes significantly, that is, the preset convergence condition is met; the convergence condition is that the moving distance of the cluster center is less than a certain threshold, or the number of iterations reaches a predetermined value;

[0017] Furthermore, in S3, the starting point of the cable extension is selected: in each cluster, the distances from all points to the geometric center of the cluster are calculated; then, the actual point closest to the geometric center is selected from these points and determined as the starting point of the cable extension; this starting point will be used as the starting position for the subsequent cluster segmentation of cables and dampers based on the density point cloud algorithm.

[0018] Furthermore, in S4, starting from the starting point of the cable extension, a density-based clustering algorithm is used to classify a certain number of points within a radius threshold around the starting point into one category, and expand to the surrounding area until the vertical coordinate of the point cloud reaches the set threshold range and stops expanding. At this time, the point cloud should include the cable and the damper connected to it.

[0019] Furthermore, in S5, the process of analyzing and obtaining the spatial straight line equation of each cable is as follows: collecting the clustering result data of cable slices at different heights, then fitting the spatial straight line equation, and performing spatial straight line fitting using the least squares method.

[0020] Furthermore, in S5, all point clouds with the cable space straight line equation as the reference and the cable radius as the threshold are deleted from the point cloud obtained in S4, specifically: traverse the point cloud obtained in S4, calculate the distance from the point to the cable space straight line, and judge and delete the point cloud.

[0021] Furthermore, in S6, the spatial straight line equation of the cable is obtained, and the set height is selected and substituted into the position coordinates of the cable anchor end, and the robot's design reserved space is added to obtain the robot's task point coordinates a; after obtaining the point cloud file of only the damper, the position coordinates b of the damper top are obtained through point cloud clustering based on the same principle as S3.

[0022] Compared with the prior art, the present invention has the following advantages:

[0023] (1) High efficiency:

[0024] Automatic planning of task points: This method can automatically determine the robot's detection task point coordinates and acquisition angles through a series of rigorous point cloud processing steps, such as clustering and equation analysis, without manual intervention and a large number of measurement calculations. This greatly saves planning time and improves the efficiency of task planning, allowing the robot to be put into the work of abnormal cable damper connection detection more quickly.

[0025] Fast data processing: With the help of advanced algorithms, such as the FAST_LIO 3D SLAM algorithm used to obtain the scene point cloud of the bridge deck inspection road, and the normal estimation method based on random sampling consistency (RANSAC) to process point cloud data, it is possible to quickly and accurately process a large amount of 3D point cloud information, effectively shortening the entire cycle from data collection to completion of task point planning.

[0026] (2) Accuracy

[0027] Accurate point cloud analysis: In the process of point cloud clustering and segmentation, the k_means clustering method and density-based clustering algorithm are used to accurately distinguish the point clouds of cables and dampers. At the same time, the spatial linear equation of the cables is analyzed by clustering the results of cable slices at different heights, which further improves the accurate grasp of the spatial positions of cables and dampers, and provides a reliable data basis for the calculation of subsequent task point coordinates and acquisition angles.

[0028] Accurate task point positioning: The task point coordinates are determined by combining the cable anchor end position and the reserved space in the robot design, and the top coordinates of the damper are accurately obtained through point cloud clustering, ensuring that the robot can accurately reach the detection position and collect data at a suitable angle, thereby improving the accuracy of the detection results.

[0029] (3) Reliability

[0030] Data integrity assurance: When acquiring point cloud data, all cable point clouds are retained by selecting the height above the street lamp for slicing, avoiding data loss and omission, and ensuring data integrity for subsequent analysis and processing. In the process of cable point cloud stripping, the cable point cloud can be effectively removed to obtain a pure damper point cloud file, further improving the reliability of the data.

[0031] Stable algorithm support: The various algorithms used, such as the FAST_LIO 3D SLAM algorithm, the RANSAC normal estimation method, the k_means clustering algorithm, etc., are mature algorithms that have been verified in practice. They have high stability and reliability and can run stably in different scenarios and conditions to ensure the accuracy and reliability of mission point planning.

[0032] (4) Adaptability

[0033] Complex scene adaptation: This method can be used in complex bridge deck inspection road scenes, effectively filtering out data far away from the area of ​​interest, and accurately extracting information about cables and dampers from a large amount of point cloud data. Even in the presence of noise, interference, or irregular geometric shapes, the target object can be accurately identified through clustering and segmentation algorithms, adapting to the detection needs of different bridge structures and environmental conditions.

[0034] Flexible task adjustment: Since this method is based on the processing and analysis of point cloud data, when the bridge structure changes or the detection task requirements are adjusted, it is only necessary to re-acquire the point cloud data and perform corresponding processing to quickly generate a new task point plan, which has strong flexibility and adaptability.

[0035] (5) Cost-effectiveness

[0036] Reduce labor costs: Automated task point planning reduces the degree of manual participation, reduces dependence on professionals, and reduces the manpower required for manual measurement and planning, thereby reducing inspection costs.

[0037] Improve equipment utilization: Accurate task point planning enables the robot to complete inspection tasks more efficiently, reduces unnecessary movement and operation, improves the robot's work efficiency and equipment utilization, and further reduces inspection costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 A flow chart of the automatic planning method for robot detection task points of cable damper connection anomalies;

[0039] Figure 2 Provide an axonometric view of the hardware device;

[0040] Figure 3 This is a schematic diagram of the bridge deck maintenance road scene;

[0041] Figure 4 This is a schematic diagram of the robot in the bridge deck maintenance lane;

[0042] Figure 5 A schematic diagram of the operator controlling the robot;

[0043] Figure 6 This is a schematic diagram of the point cloud scene of the bridge maintenance road;

[0044] Figure 7 It is a schematic diagram of all cables after high-level slicing;

[0045] Figure 8 Schematic diagram of the cable extension point (green);

[0046] Fig. 9 Schematic diagram of clustered backstays and damping;

[0047] Fig.10 is a schematic diagram of the damper point cloud;

[0048] Fig.11 Schematic diagram of the coordinates a of the task point and the coordinates b of the damping top;

[0049] Fig.12 Schematic diagram for determining the acquisition angle.

[0050] Figure numerals: 1-perception module; 2-communication module; 3-control module; 4-motion module. DETAILED DESCRIPTION

[0051] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. Component models, material names, connection structures, control methods, algorithms and other features not clearly described in this technical solution are all considered to be common technical features disclosed in the prior art.

[0052] Example 1

[0053] This embodiment provides a method for automatically planning a robot task point for detecting abnormal cable damper connection. Figure 1 As shown, point cloud clustering and pose calculation include the following steps:

[0054] S1: Bridge deck inspection road scene point cloud acquisition: Figure 3 , 4 As shown in FIG5 , through the FAST_LIO 3D SLAM algorithm embedded in the hardware device, the operator independently remotely controls the hardware device to move along the entire length of the bridge deck inspection road. After completion, the pcd file is saved to obtain a 3D point cloud scene file of the bridge deck inspection road. The 3D point cloud of the bridge deck inspection road scene is shown in FIG5 . Figure 6 shown.

[0055] S2: Obtaining cable slices at a specific height: Based on the normal estimation method of random sampling consensus (RANSAC), filter the data, estimate the normal, separate the model, and select the slices at the height above the street lamp to retain the cable point cloud;

[0056] S3: Clustering of cable extension starting points: The point cloud clusters near the same height of each cable are clustered by the k_means clustering method, and an actual point closest to its geometric center is used as the cable extension starting point; Figure 8 shown.

[0057] S4: Cluster segmentation of cables and dampers based on density point cloud algorithm: Starting from the starting point of the cable extension, based on the density clustering algorithm, the point cloud is expanded to include cables and dampers; Fig. 9 shown.

[0058] S5: Cable point cloud stripping: The spatial linear equation of each cable is obtained by analyzing the cable slice clustering results at different heights in S3. All point clouds with the spatial linear equation of each cable as the reference and the cable radius as the threshold are deleted from the point cloud obtained in S4, and then a point cloud file with only the damper is obtained; the point cloud file of the damper is as follows: Fig.10 shown.

[0059] S6: Obtaining the coordinates of the task point and the top coordinates of the damper: The coordinates of the anchor end are obtained from the linear equation of the cable space, and the coordinates of the task point a are obtained by combining the reserved space; the coordinates of the top of the damper b are obtained by clustering the damper point cloud file; Fig.11 shown.

[0060] S7: Robot collection angle calculation: Determine the unique spatial straight line equation through a and b, and obtain the angle between it and the three axes in the point cloud file coordinate system to obtain the robot's horizontal rotation angle and pitch angle at this point, thereby realizing automatic planning of detection task points. Fig.12 shown.

[0061] In a specific embodiment, in S1, Figure 2 As shown, the hardware device includes: a perception module 1, a motion module 4, a control module 3 and a communication module 2.

[0062] In a specific implementation, the perception module 1 is a 3D laser radar with a built-in IMU, which is located at the top of the device; the control module 3 includes a control mainboard and a switch, which are located in the upper shell, and an optical camera is also provided in the shell; the communication module 2 is a built-in traffic card router, which can realize data transmission; the motion module 4 is located at the bottom and is a wheeled chassis.

[0063] In a specific implementation, in S2, the normal estimation method based on random sampling consensus (RANSAC) is specifically as follows: filtering out data far away from the area of ​​interest; estimating the surface normal of each point; separating the plane model and the cylindrical model from the point cloud using the normal estimation method; and selecting the height above the street lamp for slicing to retain and save all the cable point clouds. Figure 7 shown.

[0064] In a specific implementation, in S3, the cluster convergence judgment is: repeating the above steps of distance calculation, cluster allocation and cluster center update until the cluster center no longer changes significantly, that is, the preset convergence condition is met; the convergence condition is that the moving distance of the cluster center is less than a certain threshold, or the number of iterations reaches a predetermined value;

[0065] In a specific implementation, in S3, the starting point of the cable extension is selected: in each cluster, the distances from all points to the geometric center of the cluster are calculated; then, the actual point closest to the geometric center is selected from these points and determined as the starting point of the cable extension; this starting point will be used as the starting position for the subsequent cluster segmentation of cables and dampers based on the density point cloud algorithm.

[0066] In a specific implementation, in S4, starting from the starting point of the cable extension, a density-based clustering algorithm is used to classify a certain number of points within a radius threshold around the starting point into one category, and expand to the surrounding area until the vertical coordinate of the point cloud reaches the set threshold range and stops expanding. At this time, the point cloud should include the cable and the damper connected to it.

[0067] In a specific implementation, in S5, the process of analyzing and obtaining the spatial straight line equation of each cable is: collecting the clustering result data of cable slices at different heights, then fitting the spatial straight line equation, and performing spatial straight line fitting using the least squares method.

[0068] In a specific implementation, in S5, all point clouds with the cable space straight line equation as the reference and the cable radius as the threshold are deleted from the point cloud obtained in S4, specifically: traverse the point cloud obtained in S4, calculate the distance from the point to the cable space straight line, and judge and delete the point cloud.

[0069] In a specific implementation, in S6, the spatial straight line equation of the cable is obtained, the set height is selected, and the position coordinates of the cable anchor end are substituted and added to the design reserved space of the robot to obtain the robot task point coordinates a; after obtaining the point cloud file of only the damper, the position coordinates b of the damping top are obtained through point cloud clustering based on the same principle as S3.

[0070] Components not described in detail in this embodiment are all existing components that can be purchased through public channels.

[0071] The above description of the embodiments is to facilitate the understanding and use of the invention by those skilled in the art. It is obvious that those skilled in the art can easily make various modifications to these embodiments and apply the general principles described herein to other embodiments without creative work. Therefore, the present invention is not limited to the above embodiments, and improvements and modifications made by those skilled in the art based on the disclosure of the present invention without departing from the scope of the present invention should be within the scope of protection of the present invention.

Claims

1. A method for automatically planning task points for robot detection of abnormal cable damper connection, characterized in that: Including point cloud clustering and pose calculation, including the following steps: S1: Obtaining the scene point cloud of the bridge deck inspection road: Through the FAST_LIO 3D SLAM algorithm embedded in the hardware device, the operator independently remotely controls the hardware device to move along the entire length of the bridge deck inspection road. After completion, the pcd file is saved to obtain the 3D point cloud scene file of the bridge deck inspection road; S2: Obtaining cable slices at a specific height: Based on the normal estimation method based on random sampling consistency, filter the data, estimate the normal, separate the model, and select the slices at the height above the street lamp to retain the cable point cloud; S3: Clustering of cable extension starting points: The point cloud clusters near the same height of each cable are clustered by the k_means clustering method, and an actual point closest to its geometric center is used as the cable extension starting point; S4: Cluster segmentation of cables and dampers based on density point cloud algorithm: Starting from the starting point of cable extension, based on the density clustering algorithm, the point cloud is expanded to include cables and dampers; S5: Cable point cloud stripping: The spatial linear equation of each cable is obtained by analyzing the cable slice clustering results at different heights in S3. All point clouds with the spatial linear equation of each cable as the reference and the cable radius as the threshold are deleted from the point cloud obtained in S4, and then a point cloud file with only the damper is obtained; S6: Obtaining the coordinates of the task point and the top coordinates of the damper: The coordinates of the anchor end are obtained from the linear equation of the cable space, and the coordinates of the task point a are obtained by combining the reserved space; the coordinates of the top of the damper b are obtained by clustering the damper point cloud file; S7: Calculation of robot acquisition angle: The unique spatial straight line equation is determined by a and b, and the angle between it and the three axes in the point cloud file coordinate system is obtained to obtain the robot's horizontal rotation angle and pitch angle here, thereby realizing automatic planning of detection task points.

2. The method for automatically planning task points for robot detection of abnormal cable damper connection according to claim 1 is characterized in that: In S1, the hardware device includes: a perception module (1), a motion module (4), a control module (3) and a communication module (2).

3. The automatic planning method for robot detection task points of cable damper connection abnormality according to claim 2 is characterized in that: The perception module (1) is a 3D laser radar with a built-in IMU, which is located at the top of the device; the control module (3) includes a control mainboard and a switch, which is located in the upper shell, and an optical camera is also arranged in the shell; the communication module (2) is a built-in flow card router, which can realize data transmission; the motion module (4) is located at the bottom and is a wheeled chassis.

4. The automatic planning method for robot detection task points of cable damper connection abnormality according to claim 1 is characterized in that: In S2, the normal estimation method based on random sampling consistency is as follows: filter out data far away from the area of ​​interest; estimate the surface normal of each point; separate the plane model and the cylindrical model from the point cloud using the normal estimation method; and retain and save all the cable point clouds by selecting the height above the street lamp for slicing.

5. The automatic planning method for robot detection task points of cable damper connection abnormality according to claim 1 is characterized in that: In S3, cluster convergence judgment: repeat the above steps of distance calculation, cluster allocation and cluster center update until the cluster center no longer changes significantly, that is, the preset convergence condition is met; the convergence condition is that the moving distance of the cluster center is less than a certain threshold, or the number of iterations reaches a predetermined value.

6. The automatic planning method for robot detection task points of cable damper connection abnormality according to claim 1 is characterized in that: In S3, the starting point of the cable extension is selected: in each cluster, the distances of all points to the geometric center of the cluster are calculated; Then, the actual point closest to the geometric center is selected from these points and determined as the starting point of the cable extension; this starting point will serve as the starting position for the subsequent clustering segmentation of cables and dampers based on the density point cloud algorithm.

7. The method for automatically planning task points for robot detection of abnormal cable damper connection according to claim 1 is characterized in that: In S4, starting from the starting point of the cable extension, a density-based clustering algorithm is used to classify a certain number of points within a radius threshold around the starting point into one category and expand to the surrounding area until the vertical coordinate of the point cloud reaches the set threshold range and stops expanding. At this time, the point cloud should include the cable and the damper connected to it.

8. The method for automatically planning task points for robot detection of abnormal cable damper connection according to claim 1, characterized in that: In S5, the process of analyzing and obtaining the spatial straight line equation of each cable is as follows: collecting the clustering result data of cable slices at different heights, then fitting the spatial straight line equation, and performing spatial straight line fitting using the least squares method.

9. The automatic planning method for robot detection task points of cable damper connection abnormality according to claim 1 is characterized in that: In S5, all point clouds with the cable space straight line equation as the reference and the cable radius as the threshold are deleted from the point cloud obtained in S4. Specifically, the point cloud obtained in S4 is traversed, the distance from the point to the cable space straight line is calculated, and the point cloud is judged and deleted.

10. The automatic planning method for robot detection task points of cable damper connection abnormality according to claim 1 is characterized in that: In S6, the spatial straight line equation of the cable is obtained, and the set height is selected and substituted into the position coordinates of the cable anchor end, and the robot's design reserved space is added to obtain the robot's task point coordinates a; after obtaining the point cloud file of only the damper, the position coordinates b of the damper top are obtained through point cloud clustering based on the same principle as S3.

Images