Distribution network wire identifying and positioning system based on single-line laser radar

Abstract

The invention provides a distribution network wire identifying and positioning system based on a single-line laser radar. The single-line laser radar (21) is fixedly installed at the tail end of the axis of a second wrist joint (34) of a six-degree-of-freedom mechanical arm, and the normal vector of a scanning plane is parallel to the axis of a second wrist joint (32); the scanning dimension and position of the laser radar are dynamically adjusted through the mechanical arm, and recognition and positioning of the distribution network wire component are completed through component point cloud segmentation and operation point recognition and positioning based on geometric constraint-frequency statistical method-semantic environment-operation standard. And the operation points of the insulator and the wire can be accurately identified and accurately positioned. The system has the advantages of high efficiency and economical efficiency, can adapt to a three-dimensional scene in which electric poles and wires are staggered, provides a full-automatic solution for a distribution network live working scene, and promotes intelligent transformation of power operation and maintenance.

Description

Technical Field

[0001] The present invention belongs to the technical fields of live working on distribution networks and 3D visual perception, etc., and particularly relates to a distribution network wire identification and positioning system based on a single-line lidar. Background Art

[0002] With the gradual deepening of the construction of the distribution network, both power supply enterprises and power users have put forward higher requirements for the reliability of power supply. High-altitude manual live working has high requirements for the physical fitness of operators and there are hidden dangers of electric shock and high-altitude fall. To solve the problem of personal safety in live working, it has become an inevitable trend for intelligent robots for live working on distribution networks to replace humans to perform relatively complex tasks such as distribution network connection. At present, the locations of live working on distribution networks are distributed in different cities and suburbs, with complex distributions and diverse environmental changes. Moreover, affected by the environmental light intensity, it poses a severe challenge to the identification and positioning of slender cylindrical wires with weak texture and the realization of the autonomous perception function of intelligent robots for live working on distribution networks.

[0003] The existing identification and positioning of energized distribution network conductors mainly fall into two categories: visual image recognition and three-dimensional point cloud recognition. In the field of visual image recognition and positioning, Patent CN118015032A discloses a high-precision identification and positioning method, device, and storage medium for transmission lines. By using color images and depth maps, the transmission lines are fitted and positioned based on the normal slope of the left and right edge curves of the transmission lines. However, the camera is sensitive to changes in outdoor light intensity, resulting in the target identification and positioning being easily affected by light, which greatly limits the applicability of this method. In terms of three-dimensional point cloud recognition and positioning, Patent CN117036826A discloses a method for identifying and positioning transmission lines in energized distribution network operations. Multiple three-dimensional lidar are used for more comprehensive data collection, and the ability to describe the characteristics of transmission lines is relatively strong. Combining with an SVM classifier and correlation analysis, the recognition accuracy is quite high. However, the cost is high, the system is complex, the requirements for hardware devices are harsh, and it can only identify transmission lines, with a relatively single application scenario. Patent CN114627374A discloses a point cloud acquisition system and insulator identification and positioning method based on lidar and a pan-tilt head. The lidar is vertically installed at the center of the pan-tilt head to increase the scanning range. Patent CN119340847A discloses a lightweight live working robot system and working method based on an insulated boom truck, with a fixed radar installation position. The above two solutions have blind spots in the scanning space and cannot flexibly change the scanning position. Patent CN118093706A proposes a live working robot, system, and working method for distribution networks. According to the characteristics of different targets (leads, lead ends, running lines), multiple recognition methods such as depth camera point cloud acquisition, binocular camera video image acquisition, and lidar scanning are matched, which improves the recognition accuracy of targets in complex environments to a certain extent. However, the equipment cost is high and the hardware dependence is high. Patent CN116883606A discloses a three-dimensional model construction method and system. The clustered point cloud clusters are geometrically model-fitted to decompose different components, the connection topological relationship between components is analyzed and determined, a small local three-dimensional reconstruction is performed using a model library, and then the small pieces are combined to form a complete three-dimensional model. Although the modeling of insulators and conductors is mentioned, its method relies on a general model library and geometric fitting, and the use of the model library requires the premise that the point cloud has obvious geometric appearance characteristics for fitting. The problem of identifying and positioning irregular or low-density point clouds is not mentioned, lacking a refined identification and positioning method specific to the power scenario. Summary of the Invention

[0004] In view of this, aiming at the defects and deficiencies existing in the prior art, the purpose of the present invention is to provide a distribution network wire identification and positioning system based on a single-line lidar, focusing on the refined identification and positioning in the power scenario, embedding industry standards into the algorithm, and realizing a closed loop from reconstruction to operation guidance, so as to effectively solve the problems of high-precision and automatic identification and positioning of wires and distribution network components with irregular or low-density point clouds in complex distribution environments and variable conditions in the prior art, effectively improving the hardware utilization and operation efficiency, and reducing equipment switching and operation steps.

[0005] The solution of the present invention aims to improve the automatic operation and maintenance level of power facilities through high-precision three-dimensional perception technology. The single-line lidar is installed at the position of the wrist joint 2 of the six-degree-of-freedom collaborative robotic arm. By dynamically adjusting the scanning dimension and position of the lidar through the robotic arm, the vision limitation of traditional fixed scanning is broken, and the scanning range is significantly expanded. At the same time, when the robotic arm executes the scanning and modeling task, it can cooperate to perform the grasping task of the operation tool, greatly improving the operation efficiency and accuracy. The system adopts depth-intensity dual-modal filtering, three-dimensional reconstruction and point cloud processing methods, combines the K-D tree spatial index, improved voxel grid downsampling, DBSCAN and PCA collaborative high-voltage line component point cloud segmentation technology, and operation point identification and positioning based on geometric constraints-frequency statistics method-semantic environment-operation standards to accurately identify and accurately position the operation points of insulators and wires. The system has the advantages of high efficiency and economy, can adapt to the three-dimensional scene of poles and wires intersecting, provides a full-automatic solution for the live working scenario of the distribution network, and promotes the intelligent transformation of power operation and maintenance.

[0006] The specific technical solution adopted by the present invention to solve its technical problems is: A distribution network wire identification and positioning system based on a single-line lidar: The single-line lidar (21) is fixedly installed at the axis end of the wrist joint two (34) of the six-degree-of-freedom robotic arm, and the normal vector of the scanning plane is parallel to the axis of the wrist joint two (32); by dynamically adjusting the scanning dimension and position of the lidar through the robotic arm, the identification and positioning of the distribution network wire components are completed by using component point cloud segmentation and operation point identification and positioning based on geometric constraints-frequency statistics method-semantic environment-operation standards.

[0007] Further, a rigid bracket (22) is connected to a single-line lidar (21). The upper bracket cover (24) and the lower bracket cover (23) of the symmetric structure of the rigid bracket (22) wrap the second wrist joint (34), and the second wrist joint (34) is fixed to the rotational positions of the first wrist joint (33) and the third wrist joint (35) in a coaxial manner to ensure the position and attitude of the single-line lidar (21) are fixed. The first 4 joints of the robotic arm are used to adjust the position and attitude of the single-line lidar, including three joints of the base (31), the shoulder (32), and the elbow (36) for changing the position of the single-line lidar (21) and the first wrist joint (33) for adjusting the angle of the single-line lidar (21).

[0008] Further, after the robotic arm in the retracted state initializes its movement, it grabs the end operating tool. During this process, by coordinately adjusting the attitude of the single-line lidar, point cloud scanning and modeling are synchronously performed. When the initialization is completed, the point cloud data acquisition is finished. The method of controlling the attitude of the single-line lidar is to adjust the second wrist joint (34) during the process of the end position of the robotic arm moving to the initialization end position point, so that the single-line lidar (21) maintains an upward scanning angle. When the robotic arm moves to the tool library to grab the operating tool, the algorithms for identifying and positioning high-voltage wires and insulators are synchronously executed to complete the pose estimation of the operation target point.

[0009] Further, the process of dynamically adjusting the scanning dimension and position of the lidar by the robotic arm and completing the identification and positioning of distribution network wire components by using component point cloud segmentation and operation point identification and positioning based on geometric constraint - frequency statistics method - semantic environment - operation standard includes: By controlling the attitude of the lidar and enabling the lidar to scan during the process of the robotic arm moving to the tool grabbing point, the joint angles of the robot and the point cloud data of the distribution network environment collected by the lidar are synchronously obtained; Perform point cloud trailing filtering and denoising on the point cloud data based on the dual modalities of depth and intensity; Perform 3D reconstruction on the point cloud, including pose solution and point cloud coordinate transformation; Process the 3D point cloud data to identify the target components; Perform fitting segmentation on the 3D point cloud instances, and determine the pose of the operation point according to the semantic environment and operation standard.

[0010] Further, the point cloud trailing filtering combines the changes in depth values and reflection intensity values between each point and its adjacent points. When the depth change of adjacent point clouds is greater than a certain threshold, it is judged as trailing point cloud and removed. The remaining point cloud is subjected to the removal of abnormal frequency values of the reflection intensity of adjacent points for optimization.

[0011] Further, the pose solution is based on the structural parameters of the robotic arm, the laser emission height parameter provided by the radar, and the structural parameters of the radar mounting bracket to obtain the D-H parameters of the robotic arm of the vision system. Through the kinematic modeling of the 4-link mechanism, the point cloud coordinate transformation matrix is obtained, and the point cloud data is transformed into the robotic arm coordinate system to complete the 3D point cloud stitching.

[0012] Further, the processing of the 3D point cloud data to identify the target component specifically includes the following steps: Perform a pass-through filter on the 3D point cloud data to filter out invalid point cloud data; Based on the K-D tree search model, reduce the sampling of the point cloud data by improving the voxel grid method, retaining important details while reducing the number of points in the point cloud; Execute the component point cloud segmentation in cooperation with DBSCAN-PCA; Identify and locate the operation points based on geometric constraints - frequency statistics - semantic environment - operation standards.

[0013] Further, the improved voxel grid method uses a cuboid voxel unit. For each point after voxel downsampling, the KNN algorithm in the model K-D tree is used to find the centroid point and the closest point in the original point cloud as the new point in the improved downsampled point cloud.

[0014] Further, the component point cloud segmentation in cooperation with DBSCAN-PCA is specifically as follows: Set the parameters r and k of the DBSCAN algorithm, analyze each point to determine whether it is a core point; if there are at least k points in the neighborhood of a certain point, it is marked as a core point; select a core point as the initial point of the cluster, use the K-D tree model to accelerate the neighborhood search, and gradually expand the cluster until it cannot be expanded; all points directly or indirectly connected to the core point will ultimately be classified into the same cluster; each cluster is assigned a unique label, and points not assigned to any cluster are marked as noise points; use PCA principal component analysis for each point cloud cluster to extract the first principal component; and make a judgment based on the following geometric constraint conditions: if the absolute value of the angle between the principal component and the Y-axis is less than 60 degrees and the absolute value of the angle between the principal component and the Z-axis is greater than 60 degrees, then it is determined that the cluster is the point cloud of the high-voltage line component; if the absolute value of the angle between the principal component and the Z-axis is less than 60 degrees and the absolute value of the angle between the principal component and the Y-axis is greater than 60 degrees, then it is determined that the cluster is the point cloud of the utility pole.

[0015] Further, the identification and positioning of the operation points based on geometric constraints - frequency statistics - semantic environment - operation standards are specifically as follows: A custom cylinder model is used to regularize the shape of the point cloud of the irregular insulator porcelain disk; the RANSAC line fitting algorithm is used to identify and fit the high-voltage line; the base point of the custom cylinder model is set as the point on the line fitted by RANSAC, and it rotates around the cylinder base point and the angle between the cylinder axis and the first principal component direction gradually increases evenly, with a maximum angle of 30 degrees. By statistically analyzing the distribution of intensity values and setting a threshold to filter out the intensity points with lower frequencies, the position of the point cloud area satisfying the insulator is found; through the position of the point cloud area of the insulator and the direction of the line fitted by the RANSAC algorithm, and meeting the technical requirement that the length of the insulation stripping position from the center line of the insulator is between 50 cm and 150 cm. The formula for solving the expected operation point position of the high-voltage line is as follows: Where, the position of the insulator , the center line direction and the length L from the operation point to the insulator, the position of the expected operation point of the high-voltage line .

[0016] Compared with the prior art, the beneficial effects of the present invention and its preferred solutions at least include: 1. The dynamically adjustable collaborative robot and lidar system provided by the present invention not only break through the vision limitation of fixed scanning, but also can flexibly adjust the scanning angle and range according to needs to ensure comprehensive and accurate point cloud data acquisition in complex environments; the manipulator enters the initial operation posture and the process of grasping the end operation tool from the tool library is completed by coordinately adjusting the lidar posture to perform point cloud scanning and modeling, improving the hardware utilization, reducing equipment switching and operation steps, and effectively improving the operation efficiency.

[0017] 2. The reflection intensity and depth dual-modal filtering provided by the present invention solves the problem of trailing noise in dynamic scanning; uses cuboid voxels to replace traditional cube voxels to overcome the problem of poor dynamic adaptability of cube voxels, combines the KNN algorithm to extract the nearest neighbor points in the original point cloud as the representative points after downsampling, and retains local geometric details; the DBSCAN-PCA collaborative combination and the angle constraint with the coordinate axis are used to segment the point cloud of high-voltage line components; by constructing a custom cylinder model and RANSAC wire fitting, the insulator is identified and positioned based on geometric constraints and frequency statistics, combined with the direction of the wire fitted by RANSAC, and combined with industry specifications (such as the insulation stripping position is 50 - 150 cm from the insulator) to calculate the spatial position of the wire operation point, embedding industry standards into the algorithm, and realizing a closed loop from reconstruction to operation guidance.

[0018] 3. The system provided by the present invention has a high degree of adaptability and can work efficiently in a complex three-dimensional environment with poles and wires intertwined. It is particularly suitable for the live working scenario of the distribution network and provides a fully automated solution for the intelligent operation and maintenance of power facilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The present invention will be further described in detail below in conjunction with the drawings and specific embodiments: Figure 1 Equipment diagram of the distribution network wire identification and positioning system based on a single-line lidar provided by an embodiment of the present invention; Figure 2 Installation position diagram of the single-line lidar in the visual perception module provided by an embodiment of the present invention; Figure 3 Overall schematic diagram of the distribution network wire identification and positioning system based on a single-line lidar provided by an embodiment of the present invention; Figure 4 Schematic diagram of the control function of the control system provided by an embodiment of the present invention; Figure 5 Flowchart of the operation point identification and positioning method provided by an embodiment of the present invention; Figure 6 Flowchart of the method for identifying insulators and wires by the point cloud processing algorithm provided by an embodiment of the present invention.

[0020] In the figure: 1. Human-computer interaction terminal; 2. Visual perception module; 3. Manipulator motion module; 4. Independent power supply module; 5. Dual-mode communication module; 6. Control module; 7. Insulator; 8. Cross arm; 9. Wire; 10. Pole.

[0021] 21. Single-line lidar; 22. Rigid bracket; 23. Bracket lower cover; 24. Bracket upper cover; 31. Base; 32. Shoulder; 33. Wrist joint one; 34. Wrist joint two; 35. Wrist joint three; 36. Elbow. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] To make the features and advantages of this patent more obvious and understandable, specific embodiments are given below for detailed description as follows: It should be noted that the following detailed descriptions are all illustrative and are intended to provide further explanations for the present application. Unless otherwise specified, all technical and scientific terms used in this specification have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present application belongs.

[0023] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular forms are also intended to include the plural forms. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they specify the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0024] Referring to Figures 1 - 6 , the overall structure of the equipment of the distribution network wire identification and positioning system based on a single-line lidar provided by the embodiment of the present invention is as Figure 1 shown. The distribution network wire identification and positioning system based on a single-line lidar includes a human-machine interaction terminal 1 and a live working robot; the live working robot includes a visual perception module 2, a robotic arm motion module 3, an independent power supply module 4, a dual-mode communication module 5, and a control module 6; the control module 6 is composed of an industrial computer and serves as the core control unit, and controls the visual perception module 2 and the robotic arm motion module 3 through the dual-mode communication module 5.

[0025] As Figure 2 shown, the visual perception module 2 uses a single-line lidar 21 for high-precision three-dimensional reconstruction of the distribution network environment and identification and positioning of target objects; the dual-mode communication module 5 is composed of an industrial router and adopts a wired and wireless collaborative layout, and is responsible for feeding back the status information of the control module 6, the visual perception module 2, the robotic arm motion module 3, and the independent power supply module 4 to the human-machine interaction terminal module 1 for display; the power supply module 4 is composed of a 48V battery to supply power to each module; the single-line lidar 21 is fixed through a rigid bracket structure 22-24 (including the rigid bracket 22 connecting the single-line lidar, the lower cover 23 of the robotic arm wrist joint two wrapping bracket, and the upper cover 24 of the robotic arm wrist joint two wrapping bracket) and installed at the end of the axis of the robotic arm wrist joint two 32, and the normal vector of its scanning plane is parallel to the axis of the wrist joint two 32; the robotic arm motion module 3 is composed of a collaborative robotic arm and is used to change the scanning dimension of the visual module 2 and move the operation tool to the target point, and the two functions can be completed collaboratively to improve the modeling and operation efficiency.

[0026] More specifically, as a preferred embodiment, the installation position of the single-line lidar in the visual perception module provided by the present invention is as Figure 2As shown in the figure, a part of the rigid support structure, the single-line lidar rigid support 22, is connected to the single-line lidar 21. Another part, the support upper cover 24 with a symmetric structure, and the support lower cover 23 wrap the second robotic arm wrist joint 34, fixing the second wrist joint 34 to the rotational positions of the first wrist joint 33 and the third wrist joint 35 in a coaxial manner, ensuring the position and attitude of the single-line lidar 21 are fixed. And the collaborative robotic arm 3 is a six-joint robot. The first four joints of the six-joint robot include the base 31, the shoulder 32, the elbow 36, and the first wrist joint 33, which are used to adjust the position and attitude of the single-line lidar 21. The three joints of the base 31, the shoulder 32, and the elbow 36 are mainly used to change the position of the single-line lidar 21, and the fourth joint, the first wrist joint 33, is mainly used to adjust the angle of the single-line lidar 21. The six-joint robot includes the three joints of the base 31, the shoulder 32, and the elbow 36, which are mainly used to change the position of the end of the robot, and the three joints of the first wrist joint 33, the second wrist joint 34, and the third wrist joint 35, which are mainly used to adjust the attitude of the end of the robot.

[0027] The working process of the power distribution wire identification and positioning system based on a single-line lidar provided by the present invention is as Figure 3 shown. The live working robot is lifted by an insulated boom truck to below the wire where live working is required, and then the artificial interaction terminal is operated to control the robotic arm motion module 3 and the visual perception module 2 to identify the wire 9, insulator 7, cross arm 8, and pole 10, determine the position and attitude of the working point of the wire 9, and control the robotic arm to move to the target position to complete the operation task.

[0028] Based on the design of the above system device structure, the control function of the control system provided by the embodiment of the present invention is as Figure 4 shown. The control system communicates with the robotic arm and the single-line lidar through a dual-mode communication module, controls the robotic arm to move for scanning and modeling and operating on the target point of the end flange, controls the scanning state of the single-line lidar, and simultaneously obtains the angles of each joint of the robotic arm and the point cloud data collected by the single-line lidar, and obtains three-dimensional point cloud information after being processed by the point cloud coordinate conversion algorithm.

[0029] As a preferred solution of this embodiment, the control module controls the robotic arm motion module to enter the initial operation attitude and grab the end operation tool from the tool library through dual-mode communication. During the movement to the target position, the system always pays attention to controlling the attitude of the single-line lidar and synchronously performs point cloud scanning and modeling. When the point cloud data is collected at the target position, the recognition and positioning algorithm is used to process the point cloud data to obtain the operation position and attitude, and the robotic arm returns to the initial operation attitude and then can carry the operation tool to move to the target operation position.

[0030] More specifically, when controlling the attitude of the single-line lidar during the process of moving the end position of the robotic arm to the target point, the second wrist joint of the robotic arm is adjusted at all times to keep the lidar at an upward scanning angle. When entering the range above or below the target point, the end attitude is adjusted for precise alignment to grab the tool.

[0031] On this basis, as Figure 5 shown, the operation point identification and positioning method of the distribution network wire identification and positioning system based on a single-line lidar provided in this embodiment includes the following steps: Step 1: During the process of moving the robotic arm to the tool grasping point, control the attitude of the lidar and enable the lidar to scan, and synchronously obtain the robot joint angles and the point cloud data of the distribution network environment collected by the lidar; Step 2: Perform point cloud trailing filtering and denoising on the point cloud data based on depth and intensity characteristics; Step 3: Perform three-dimensional reconstruction on the point cloud. The three-dimensional reconstruction includes pose solution and point cloud coordinate transformation; Step 4: Use a point cloud processing algorithm to identify insulators and wires from the three-dimensional point cloud data; Step 5: Perform fitting and segmentation on the three-dimensional point cloud instances, and determine the pose of the operation point according to the semantic environment and operation standards.

[0032] As a preferred solution of this embodiment, in Step 2, an innovative dual-modal trailing filtering method is adopted. By combining the changes in depth values and reflection intensity values between each point and its adjacent points, when the depth change of adjacent point clouds is greater than a certain threshold, it is judged as trailing point clouds and removed. For the remaining point clouds, abnormal values of the reflection intensity frequency of adjacent points are removed for optimization. In this solution, only the dual-modal data collected by a single sensor is combined with this method, which reduces the hardware complexity and cost. And through adjacent point relationship filtering, the sensitivity to filter out trailing noise points of thin-section wires is improved. The sensor is installed on the robotic arm to avoid the problem of insufficient spatial coverage of a fixed-site lidar.

[0033] As a preferred solution of this embodiment, in Step 3, for pose solution, based on the robotic arm structure parameters, the laser emission height parameter provided by the lidar, and the radar mounting bracket structure parameters, the D-H parameters of the vision system robotic arm are obtained. Through the above 4-link kinematic modeling, the point cloud coordinate transformation matrix is obtained, and the point cloud data is transformed into the robotic arm coordinate system to complete the three-dimensional point cloud stitching. This solution can significantly reduce the number of point cloud registration iterations through robotic arm kinematic constraints.

[0034] As a preferred solution of this embodiment, in Step 4, the steps of identifying insulators and wires by the point cloud processing algorithm include the following, as Figure 6 shown: Step 41: Perform pass-through filtering on the 3D point cloud data to filter out invalid point cloud data; Step 42: Establish a K-D tree search model to shorten the running time of the algorithm and improve efficiency; Step 43: Improve the voxel grid method for point cloud downsampling to reduce the number of point clouds while retaining important details; Step 44: DBSCAN and PCA collaborative point cloud segmentation technology for high-voltage line components; Step 45: Identification and positioning of operation points based on geometric constraints - frequency statistics - semantic environment - operation standards.

[0035] In a preferred embodiment, the improved voxel grid method uses cuboid voxel units instead of cube voxel units. For each point after voxel downsampling, the KNN algorithm in the model K-D tree is used to find the centroid point and the nearest point in the original point cloud as the new point in the downsampled point cloud. This solution significantly improves the accuracy and fidelity of the downsampled point cloud while reducing the error introduced by traditional methods.

[0036] In a preferred embodiment, in Step 44, the specific process of the DBSCAN and PCA collaborative point cloud segmentation technology for high-voltage line components is as follows: First, set the parameters r and k of the DBSCAN algorithm, analyze each point to determine whether it is a core point. If there are at least k points in the neighborhood of a certain point, it is marked as a core point. Subsequently, select a core point as the initial point of the cluster, use the K-D tree model to accelerate the neighborhood search, and gradually expand the cluster until it cannot be expanded. All points directly or indirectly connected to the core point will ultimately be grouped into the same cluster. Then, each cluster is assigned a unique label, and points not assigned to any cluster are marked as noise points. Finally, for each point cloud cluster, use the PCA principal component analysis method to extract its first principal component. According to the geometric characteristics of the high-voltage line, combined with the direction information of the principal component, set specific geometric constraint conditions: (1) If the absolute value of the angle between the principal component and the Y-axis is less than 60 degrees and the absolute value of the angle between the principal component and the Z-axis is greater than 60 degrees, then determine that this cluster is the point cloud of the high-voltage line component; (2) If the absolute value of the angle between the principal component and the Z-axis is less than 60 degrees and the absolute value of the angle between the principal component and the Y-axis is greater than 60 degrees, then determine that this cluster is the point cloud of the pole.

[0037] In a preferred embodiment, in Step 45, the specific process of job point recognition and positioning based on geometric constraints - frequency statistics method - semantic environment - operation standards is as follows: First, a custom cylindrical model is used to regularize the point cloud of the irregular insulator porcelain plate; then, the RANSAC line fitting algorithm is used to identify and fit the high-voltage line; then, the base point of the custom cylindrical model is set as the point on the line fitted by RANSAC, and it rotates around this cylindrical base point with the angle between the cylindrical axis and the first principal component direction increasing gradually and evenly, with a maximum angle of 30 degrees. By statistically analyzing the distribution of intensity values and setting a threshold, the intensity points with lower frequencies are filtered out, and the more representative point cloud in the data is retained, and the interference points of connected objects are removed to find the position that satisfies the insulator point cloud area. Based on the position of the insulator point cloud area and the direction of the line fitted by the RANSAC algorithm, and meeting the technical requirement that the length of the insulation stripping position from the center line of the insulator should be between 50 cm and 150 cm, the formula for solving the expected job point position of the high-voltage line is as follows: Wherein, the position of the insulator , the center line direction and the length L of the job point from the insulator, the position of the expected job point of the high-voltage line .

[0038] It should be noted that unless otherwise defined, the technical terms or scientific terms used in the present invention should have the ordinary meanings understood by those with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar terms used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. The terms such as "including" or "comprising" mean that the elements or objects appearing before this word cover the elements or objects listed after this word and their equivalents, without excluding other elements or objects. The terms such as "connected" or "linked" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The terms such as "up", "down", "left", "right" are only used to represent relative position relationships, and when the absolute position of the object being described changes, the relative position relationship may also change accordingly.

[0039] The above is only a preferred embodiment of the present invention, and it is not a limitation of the present invention in other forms. Any person skilled in the art may use the technical content disclosed above to make changes or modifications into equivalent embodiments with equivalent changes. However, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the technical solution content of the present invention still fall within the protection scope of the technical solution of the present invention.

[0040] This patent is not limited to the above-mentioned best implementation mode. Anyone inspired by this patent can obtain various other forms of a distribution network wire identification and positioning system based on a single-line lidar. All equal changes and modifications made according to the scope of the patent application of the present invention shall fall within the scope covered by this patent.

Claims

1. A distribution network conductor identification and positioning system based on a single-line laser radar, characterized in that: The single-line laser radar (21) is fixedly mounted at the end of the axis of the second wrist joint (34) of the six-degree-of-freedom robotic arm, and the normal vector of the scanning plane is parallel to the axis of the second wrist joint (32); the scanning dimension and position of the laser radar are dynamically adjusted by the robotic arm, and the identification and positioning of the distribution network conductor components are completed by using component point cloud segmentation and operation point identification and positioning based on geometric constraints-frequency statistics method-semantic environment-operation standards.

2. According to claim 1, a distribution network wire identification and positioning system based on a single-line laser radar is characterized in that: The rigid support (22) is connected to the single-line laser radar (21); the support upper cover (24) and the support lower cover (23) of the symmetrical structure of the rigid support (22) wrap the wrist joint 2 (34); the wrist joint 2 (34) is fixed in a coaxial manner with the rotational position of the wrist joint 1 (33) and the wrist joint 3 (35), respectively, so as to ensure that the position and posture of the single-line laser radar (21) are fixed; the first four joints of the mechanical arm are used to adjust the posture of the single-line laser radar, including the base (31) for changing the position of the single-line laser radar (21), the shoulder (32) and the elbow (36) three joints and the wrist joint 1 (33) for adjusting the angle of the single-line laser radar (21).

3. According to claim 1, a distribution network wire identification and positioning system based on a single-line laser radar is characterized in that: The folded state robot arm grasps the end operation tool after initialization movement. During the process, the posture of the single-line laser radar is coordinated and adjusted, and point cloud scanning modeling is synchronously performed. When the initialization is completed, the point cloud data collection is completed. The method of controlling the posture of the single-line laser radar is to adjust the wrist joint 2 (34) during the process of the robot arm end position moving to the initialization end position point so that the single-line laser radar (21) maintains an upward scanning angle; when the robot arm moves to the tool library to grasp the operation tool, the algorithm for identifying and locating the high-voltage line and the insulator is synchronously executed to complete the estimation of the operation target point posture.

4. According to claim 1, a distribution network wire identification and positioning system based on a single-line laser radar is characterized in that: The process of dynamically adjusting the scanning dimension and position of the laser radar by the mechanical arm, using the component point cloud segmentation and the operation point identification and positioning based on geometric constraints-frequency statistics method-semantic environment-operation standards to complete the identification and positioning of the distribution network conductor components includes: By controlling the posture of the laser radar and enabling the laser radar scanning when the robot arm moves to the tool grasping point, the robot joint angle and the distribution network environment point cloud data collected by the laser radar are synchronously obtained; Perform point cloud tailing filtering and denoising based on depth and intensity dual modes on point cloud data; Perform 3D reconstruction of point clouds including pose solution and point cloud coordinate transformation; Processing 3D point cloud data to identify target parts; Fit and segment the 3D point cloud instances, and determine the position and pose of the operating points based on the semantic environment and operating standards.

5. According to claim 4, a distribution network conductor identification and positioning system based on a single-line laser radar is characterized in that: The point cloud tailing filter combines the changes in depth value and reflection intensity value between each point and its adjacent points. When the depth change of adjacent point clouds is greater than a certain threshold, it is judged as a tailing point cloud and removed. The remaining point cloud removes abnormal reflection intensity frequency values ​​of adjacent points for optimization.

6. A distribution network wire identification and positioning system based on a single-line laser radar according to claim 4, characterized in that: The posture solution is based on the structural parameters of the robotic arm, the laser emission height parameters provided by the radar, and the structural parameters of the radar mounting bracket to obtain the DH parameters of the visual system robotic arm. The point cloud coordinate conversion matrix is ​​obtained through 4-link kinematic modeling, and the point cloud data is transformed into the robotic arm coordinate system to complete the three-dimensional point cloud splicing.

7. According to claim 4, a distribution network wire identification and positioning system based on a single-line laser radar is characterized in that: The processing of the three-dimensional point cloud data to identify the target component specifically comprises the following steps: Perform direct filtering on the three-dimensional point cloud data to filter out invalid point cloud data; Based on the KD tree search model, the point cloud data downsampling is improved by the voxel grid method to reduce the number of point clouds while retaining important details; Perform DBSCAN-PCA collaborative component point cloud segmentation; Identify and locate work points based on geometric constraints, frequency statistics, semantic environment and work standards.

8. The distribution network conductor identification and positioning system based on a single-line laser radar according to claim 7 is characterized in that: The improved voxel grid method uses a cuboid voxel unit, and for each point after voxel downsampling, uses the KNN algorithm in the model K-Dtree to find the nearest point between the centroid and the original point cloud as a new point in the improved downsampling point cloud.

9. The distribution network conductor identification and positioning system based on a single-line laser radar according to claim 7 is characterized in that: The DBSCAN-PCA collaborative component point cloud segmentation is as follows: set the parameters r and k of the DBSCAN algorithm, analyze each point and determine whether it is a core point; if a point contains at least k points in its neighborhood, mark it as a core point; select a core point as the initial point of the cluster, use the KD tree model to accelerate the neighborhood search, and gradually expand the cluster until it cannot be expanded; All points directly or indirectly connected to the core point will eventually be classified into the same cluster; each cluster is assigned a unique label, and points that are not classified into any cluster are marked as noise points; Use PCA principal component analysis for each point cloud cluster to extract the first principal component; The judgment is made based on the following geometric constraints: if the absolute value of the angle between the principal component and the Y axis is less than 60 degrees and the absolute value of the angle between the principal component and the Z axis is greater than 60 degrees, then the cluster is judged to be a high-voltage line component point cloud; If the absolute value of the angle between the principal component and the Z axis is less than 60 degrees and the absolute value of the angle between the principal component and the Y axis is greater than 60 degrees, the cluster is determined to be a pole point cloud.

10. A distribution network wire identification and positioning system based on a single-line laser radar according to claim 7, characterized in that: The identification and positioning of the operation points based on geometric constraints-frequency statistics-semantic environment-operation standards are specifically as follows: the irregular insulator porcelain plate point cloud is regularized by using a custom cylindrical model; the RANSAC straight line fitting algorithm is used to identify and fit the high-voltage line; the base point of the custom cylindrical model is set as a point on the straight line fitted by RANSAC, and the cylinder base point is rotated around the cylinder base point, and the angle between the cylinder axis and the first principal component direction is gradually and evenly increased, with a maximum angle of 30 degrees. The distribution of the statistical intensity value is used and a threshold is set to filter out the intensity points with lower frequency, and the position that meets the insulator point cloud area is found; the point cloud area position of the insulator and the straight line direction fitted by the RANSAC algorithm are used, and the technical requirement that the length of the insulation skin stripping position from the insulator center line is between 50cm and 150cm is met; The formula for solving the expected operating point position of the high-voltage line is as follows: The position of the insulator , centerline direction The distance L between the insulator and the operating point, the location of the expected operating point of the high-voltage line .

Images