Bolt miner roller repairing tooth holder positioning method based on three-dimensional point cloud data processing

Abstract

The invention relates to the technical field of coal mining, and discloses a three-dimensional point cloud data processing-based bolter miner roller repair toothholder positioning method, which comprises the following steps of: in the process of maintaining a bolter miner roller by using an industrial robot, acquiring three-dimensional point cloud data of an irregular surface left after cutting a damaged toothholder of a roller body through a visual sensor; a cushion block is positioned through an algorithm, the conversion relation between the posture of a tooth holder attached to the surface of the cushion block and the posture of the cushion block is calculated through the posture of a recognition point-teaching point, the tooth holder arc face is accurately attached to the surface of the cylinder cushion block, and the method comprises the following steps that S1, point cloud data are obtained; s2, preprocessing the original point cloud; s3, searching for the maximum connected domain; s4, determining a 2D center; s5, extracting point clouds in the region; s6, performing curvature segmentation; s7, acquiring a surface point cloud of the cushion block; and S8, calculating the surface pose of the cushion block. Automatic positioning of the position of the cushion block can be achieved in a visual positioning mode, and the tooth holder positioning efficiency and precision are improved.

Description

Technical Field

[0001] The invention relates to the technical field of coal mining, and in particular to a method for positioning a tooth seat for repairing a drum of an anchor digger based on three-dimensional point cloud data processing. Background Art

[0002] The bolter and miner drum is an important component in the coal mining process. It uses the cutting drum on the drum to peel coal from the coal seam through its own rotation. During use, due to the interaction with the coal seam and impurities such as stones mixed in the coal seam, the cutting drum tooth seat is severely worn under the action of long-term alternating loads, resulting in a decrease in the excavation capacity of the bolter and miner and a decrease in coal production. Considering the high cost of the drum itself, the severely worn cutting drum needs to be replaced.

[0003] The current repair method of manually positioning the gear seat is easily affected by human factors, resulting in insufficient positioning accuracy and low efficiency during the repair process, which makes it difficult to meet the high precision and high efficiency requirements of modern industry. Summary of the invention

[0004] In view of the shortcomings of the prior art, the present invention provides a method for positioning the tooth seat of an anchor mining machine drum repair based on three-dimensional point cloud data processing, which solves the problem that the current repair method using manual positioning of the tooth seat has high requirements on the technical level of workers, poor positioning accuracy and low efficiency.

[0005] To achieve the above purpose, the present invention is implemented through the following technical solutions: a method for positioning the tooth seat of a repairing anchor mining drum based on three-dimensional point cloud data processing, in the process of using an industrial robot to repair the anchor mining drum, the irregular surface left after the damaged tooth seat of the cylinder is cut is obtained through a visual sensor, the pad is positioned through an algorithm, and the conversion relationship between the posture of the tooth seat fitting the pad surface and the posture of the pad is calculated through the recognition point-teaching point posture, and the arc surface of the tooth seat is accurately fitted to the surface of the cylinder pad, including the following steps:

[0006] S1. Obtain point cloud data;

[0007] S2, preprocessing the original point cloud;

[0008] S3, find the largest connected domain;

[0009] S4, determine the 2D center;

[0010] S5, extracting point cloud in the area;

[0011] S6, curvature segmentation;

[0012] S7, obtaining a point cloud of the pad surface;

[0013] S8, calculating the surface posture of the pad;

[0014] S9. Calculate the conversion relationship between the posture of the gear seat fitting the surface of the pad and the posture of the pad.

[0015] Preferably, in the step S1, the point cloud data is obtained by obtaining the original point cloud of the roller pad surface and the nearby roller surface through binocular vision, including setting camera exposure parameters, triggering camera image acquisition and grating light emission at the same time, and obtaining the three-dimensional point cloud of the pad and its surrounding surface through phase dephasing of the acquired image.

[0016] Preferably, in the step S2, preprocessing the original point cloud refers to downsampling and outlier removal of the original point cloud. The cube size is artificially set through the moving cube method, and the center of the cube is gradually moved to traverse the entire point cloud area. Each time it moves, the points in the cube are obtained and the average is simplified to one point to achieve downsampling. The point cloud is traversed again. If the distance from the traversed point to other points is large, the current point is considered to be noise and is removed.

[0017] Preferably, in the step S3, the maximum connected domain is found by clustering the point cloud based on Euclidean distance, dividing the point cloud into several clusters, and selecting the cluster with the largest number of points as the point cloud for the next step of analysis.

[0018] Preferably, in the step S4, the 2D center is determined by setting the Z coordinates of all points in the point cloud to 0, calculating the bounding box to obtain the point cloud center, and constructing a KDtree to search for the nearest point, looking for a point with the same ID as the 2D center of the point cloud.

[0019] Preferably, in the step S6, curvature segmentation refers to calculating the curvature of the point cloud, removing points with curvature values ​​higher than a threshold, constructing a KDTree for the point cloud, calculating the normal vector of each point, traversing the point cloud, selecting the traversed point and several points closest to it, and achieving it by calculating the angle between the normal vector of the traversed point and the adjacent points.

[0020] Preferably, in step S7, the pad surface point cloud is obtained by clustering the point cloud based on Euclidean distance, calculating the point cloud centroid for each cluster, and selecting the cluster closest to the center of the original point cloud, which is the pad surface point cloud.

[0021] Preferably, in step S8, the surface pose of the pad is calculated by calculating the OBB bounding box of the point cloud, using the upper surface center point and posture of the OBB bounding box itself to represent the surface pose of the pad, so as to quickly calculate the irregular surface pose.

[0022] Preferably, in the step S9, calculating the conversion relationship between the posture of the gear seat fitting the pad surface and the posture of the pad means converting the identified posture of the pad in the robot coordinate system into a matrix expression, inverting it through LU decomposition, and multiplying it with the matrix expression of the robot posture when the gear seat is accurately fitted, so as to obtain the conversion relationship between the posture of the gear seat fitting the pad surface and the posture of the pad.

[0023] Preferably, a tooth seat positioning device for repairing an anchor mining machine drum based on three-dimensional point cloud data processing comprises a cutting drum body and a multi-axis motion mechanism, wherein a pad is arranged on the outside of the cutting drum body, a tooth seat is arranged on the outside of the cutting drum body, a tooth seat plug hole is arranged on the inside of the tooth seat, a tooth seat arc surface is designed on the surface of the tooth seat, and a binocular vision sensor is arranged on the outside of the multi-axis motion mechanism.

[0024] The present invention has at least the following beneficial effects:

[0025] 1. The position of the pad can be automatically positioned through visual positioning, improving the efficiency and accuracy of the gear seat positioning.

[0026] 2. The extraction of the pad surface through curvature segmentation has high robustness and can realize the positioning of gear seats with different inclination angles.

[0027] 3. By using the pose of the OBB bounding box to represent the pose of the irregular surface of the pad, the pad can be calculated quickly and accurately.

[0028] 4. By identifying the point-focus, the conversion relationship between the posture of the gear seat fitting the pad surface and the posture of the pad can be inferred, and the posture of the multi-axis motion mechanism when the arc of the gear seat fits the pad surface can be quickly calculated.

[0029] Other advantages, objectives and features of the present invention will be embodied in part through the following description, and in part will be understood by those skilled in the art through study and practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 It is a schematic diagram of the gear seat structure of the present invention;

[0031] Figure 2 The cutting drum structure diagram of the bolter digger of the present invention;

[0032] Figure 3 It is a schematic diagram of positioning the gear holder of the multi-axis motion mechanism of the present invention;

[0033] Figure 4 The present invention is a flow chart of a method for positioning a tooth seat for repairing a bolter drum based on three-dimensional point cloud data processing.

[0034] In the figure: 1. Gear seat socket; 2. Gear seat arc surface; 3. Gear seat; 4. Spacer; 5. Cutting drum body; 6. Multi-axis motion mechanism; 7. Binocular vision sensor. DETAILED DESCRIPTION

[0035] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0036] according to Figures 1 to 3 As shown, an embodiment of the present invention provides a positioning method for the tooth seat of a bolter drum repair based on three-dimensional point cloud data processing. In the process of using an industrial robot to repair the bolter drum, the three-dimensional point cloud data of the irregular surface left after the damaged tooth seat of the cylinder is cut is obtained by a visual sensor, the pad is positioned by an algorithm, and the conversion relationship between the posture of the tooth seat fitting the pad surface and the posture of the pad is calculated by identifying the point-teaching point posture, so that the arc surface of the tooth seat is accurately fitted to the surface of the cylinder pad, including the following steps:

[0037] S1. Obtain point cloud data;

[0038] S2, preprocessing the original point cloud;

[0039] S3, find the largest connected domain;

[0040] S4, determine the 2D center;

[0041] S5, extracting point cloud in the area;

[0042] S6, curvature segmentation;

[0043] S7, obtaining a point cloud of the pad surface;

[0044] S8, calculating the surface posture of the pad;

[0045] S9. Calculate the conversion relationship between the posture of the gear seat fitting the surface of the pad and the posture of the pad.

[0046] Specifically, obtain point cloud data: obtain the original point cloud PC of the drum pad surface and nearby drum surfaces through binocular vision ori ;

[0047] Preprocessing: PC of the original point cloud ori Perform downsampling and outlier removal to obtain point cloud PC pre ;

[0048] Find the largest connected domain: Segment the point cloud by clustering the point cloud based on Euclidean distance, and select the point cloud with the largest number of points for subsequent PC processing max ;

[0049] Determining 2D Center: Build and PC max The same point cloud but with the Z coordinate set to 0, the bounding box is calculated to get the center of the point cloud, and a KDtree is constructed to search for the nearest point. max Find the point with the same ID and record it as P mid ;

[0050] Extract point cloud in the area: mid As the center, given the length, width and height, a rectangular box is constructed, and the points in the rectangular box are extracted to form a point cloud PC rect ;

[0051] Curvature segmentation: point cloud PC rect Calculate curvature and set threshold Thresh curva , for PC rect The curvature of the points in are traversed, if the curvature of the traversed points satisfies:

[0052] Curvature>Threash curva

[0053] Then remove this traversal point, and obtain the remaining point cloud PC after the traversal is completed. left ;

[0054] Get the pad surface point cloud: After filtering out the points with larger curvature, the point cloud is divided into multiple points. left The middle point cloud is extracted, which is the point cloud PC of the center area of ​​the upper surface of the pad. mid ;

[0055] Calculate the surface pose of the pad: PC mid Calculate the OBB bounding box. The OBB bounding box, or directed bounding box, is implemented by creating a minimum circumscribed cuboid that can wrap the point cloud to determine the spatial position and size of the point cloud. The position of the bounding box itself represents the position of the pad surface, which can be expressed in a matrix as That is, the pad pose is expressed in the camera coordinate system.

[0056] Through hand-eye calibration, calculate the pose matrix of the pad in the robot coordinate system

[0057]

[0058] Through the above steps, we can identify and obtain the matrix of the pad surface pose in the camera coordinate system. The arc surface of the gear seat is accurately fitted to the surface of the pad, and the robot reads the current posture of the robot and converts it into a matrix form And calculate the pose matrix The inverse matrix Thus, the pose transformation matrix of the gear seat and the pad surface is calculated

[0059]

[0060] Finally, the tooth seat positioning calculation method during the repair process of the anchor miner drum was obtained:

[0061]

[0062] It should be noted that, if the sensor and the holding tool are not disassembled, step S9 does not need to be repeated.

[0063] In step S1, point cloud data is obtained by obtaining the original point cloud of the drum pad surface and nearby drum surfaces through binocular vision, including setting camera exposure parameters, triggering camera image acquisition and grating light emission at the same time, and obtaining the three-dimensional point cloud of the pad and its surrounding surface through phase dephasing of the acquired image.

[0064] Specifically, this step uses binocular vision technology to obtain high-precision three-dimensional point cloud data, which can accurately restore the geometric features of the pad and drum surface, laying a data foundation for subsequent positioning. By reasonably setting camera parameters and synchronous grating light, the quality and stability of data acquisition are ensured, and clear point cloud information can be obtained even in complex mining equipment environments or insufficient light conditions.

[0065] In step S2, preprocessing the original point cloud refers to downsampling and outlier removal of the original point cloud. The cube size is set artificially through the moving cube method, and the center of the cube is gradually moved to traverse the entire point cloud area. Each time it moves, the points in the cube are obtained and the average is simplified to one point to achieve downsampling. The point cloud is traversed again. If the distance from the traversed point to other points is large, the current point is considered to be noise and is removed.

[0066] Specifically, downsampling can reduce the amount of point cloud data and improve the computational efficiency of subsequent algorithms; outlier removal can effectively remove noise points to avoid their interference with curvature calculation and surface recognition. This process ensures that the point cloud data strikes a balance between data volume and accuracy, while improving the robustness of subsequent processing steps.

[0067] In step S3, the largest connected domain is found by clustering the point cloud based on Euclidean distance, dividing the point cloud into several clusters, and selecting the cluster with the largest number of points as the point cloud for the next step of analysis.

[0068] Specifically, this step uses the Euclidean distance clustering algorithm to filter out the most valuable areas from the point cloud data, avoiding the computational burden of processing the entire point cloud data. Selecting the cluster with the most points can effectively identify the main surface area of ​​the pad, thereby providing reliable input for subsequent surface positioning, further improving the efficiency and accuracy of the system.

[0069] In step S4, the 2D center is determined by setting the Z coordinates of all points in the point cloud to 0, calculating the bounding box to obtain the point cloud center, and constructing a KDtree to search for the nearest point, looking for a point with the same ID as the 2D center of the point cloud.

[0070] Specifically, by projecting the point cloud onto a two-dimensional plane and calculating the center, the data processing process can be simplified, and the KDTree efficient search algorithm can be used to quickly locate the center point of the target area. This method not only improves positioning efficiency, but also ensures accurate calculation of the two-dimensional projection center, laying the foundation for subsequent geometric matching.

[0071] In step S6, curvature segmentation refers to calculating the curvature of the point cloud, removing points with curvature values ​​higher than the threshold, constructing a KDTree for the point cloud, calculating the normal vector of each point, traversing the point cloud, selecting the traversed point and several points closest to it, and calculating the angle between the normal vector of the traversed point and the adjacent points.

[0072] Specifically, curvature calculation can identify and filter out noise points or abnormal points in high curvature areas, which is particularly suitable for processing irregular surfaces caused by cutting or wear. This step enhances the smoothness of point cloud data and provides more accurate input for subsequent bounding box calculation and pose matching.

[0073] In step S7, the pad surface point cloud is obtained by clustering the point cloud based on Euclidean distance, calculating the point cloud centroid for each cluster, and selecting the cluster closest to the center of the original point cloud, which is the pad surface point cloud.

[0074] Specifically, by clustering and calculating the centroid, the surface area of ​​the pad can be accurately located, and the target area can be quickly screened out even when the point cloud data contains interference information. This method ensures the accuracy of the pad surface point cloud extraction and provides high-quality data for subsequent pose calculations.

[0075] In step S8, the surface pose of the pad is calculated by calculating the OBB bounding box of the point cloud, using the upper surface center point and posture of the OBB bounding box itself to represent the surface pose of the pad, so as to quickly calculate the irregular surface pose.

[0076] Specifically, the OBB calculation method determines the pose of the pad surface in an efficient way, which is particularly suitable for processing complex or irregular surfaces. By utilizing the geometric characteristics of the OBB, the spatial position and orientation of the pad can be quickly determined, providing a reliable reference for the motion planning of the robot end effector.

[0077] In step S9, calculating the conversion relationship between the posture of the gear seat fitting the pad surface and the posture of the pad means converting the posture of the identified pad in the robot coordinate system into a matrix expression, inverting it through LU decomposition, and multiplying it with the matrix expression of the robot posture when the gear seat is accurately fitted, so as to obtain the conversion relationship between the posture of the gear seat fitting the pad surface and the posture of the pad.

[0078] Specifically, through matrix calculation and inference of transformation relationships, the relative positions of the gear holder and the pad in different coordinate systems can be accurately described. The introduction of LU decomposition improves computational efficiency and numerical stability, ensuring the accurate positioning of the gear holder when it fits the pad surface. This process is the key to achieving precise motion control of the robot.

[0079] The anchor mining machine drum repair tooth seat positioning device based on three-dimensional point cloud data processing includes a cutting drum body 5 and a multi-axis motion mechanism 6, a pad 4 is arranged on the outside of the cutting drum body 5, a tooth seat 3 is arranged on the outside of the cutting drum body 5, a tooth seat plug hole 1 is arranged inside the tooth seat 3, a tooth seat arc surface 2 is designed on the surface of the tooth seat 3, and a binocular vision sensor 7 is arranged on the outside of the multi-axis motion mechanism 6.

[0080] Specifically, the tooth seat 3 is an important component on the anchor mining machine drum, which is mainly used to carry and fix cutting tools (such as tooth blades). The tooth seat arc surface 2 is the part that contacts the surface of the drum. During the repair process, it is necessary to ensure that the tooth seat arc surface 2 and the surface of the pad 4 are accurately matched to restore the function of the drum. The tooth seat socket 1 is used to connect and fix the tooth seat. The pad 4 is a component used to provide a contact surface to fit with the tooth seat arc surface 2 during the repair process; its surface needs to be accurately positioned so that the tooth seat 3 can accurately dock with it. The cutting drum body 5 is the core component of the anchor mining machine. By rotating Cutting coal seams; after the tooth holder 3 is installed on the surface of the cutting drum body 5, the cutting drum body 5 will strip the coal according to the hardness and shape of the coal seam; the multi-axis motion mechanism 6 (such as an industrial robot) is responsible for accurately controlling the positioning and installation of the tooth holder; during the repair process, the multi-axis motion mechanism 6 can adjust the position of the tooth holder according to the feedback information of the sensor, and the binocular vision sensor 7 is used to obtain three-dimensional point cloud data of the pad and the surrounding area, providing high-precision positioning information; through the analysis of the point cloud data, the sensor can determine the position and posture of the tooth holder and the pad surface.

[0081] Although the embodiments of the present invention have been disclosed as above, they are not limited to the applications listed in the specification and the implementation modes, and they can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to the specific details and the illustrations shown and described herein.

Claims

1. A method for positioning the tooth seat of a bolter drum repair based on three-dimensional point cloud data processing, characterized in that: In the process of using industrial robots to repair the drum of the anchor miner, the three-dimensional point cloud data of the irregular surface left after the broken tooth seat of the cylinder is cut is obtained through the visual sensor, the pad is positioned through the algorithm, and the conversion relationship between the posture of the tooth seat fitting the pad surface and the posture of the pad is calculated through the recognition point-teaching point posture, and the arc surface of the tooth seat is accurately fitted to the surface of the cylinder pad, including the following steps: S1. Obtain point cloud data; S2, preprocessing the original point cloud; S3, find the largest connected domain; S4, determine the 2D center; S5, extracting point cloud in the area; S6, curvature segmentation; S7, obtaining a point cloud of the pad surface; S8, calculating the surface posture of the pad; S9. Calculate the conversion relationship between the posture of the gear seat fitting the surface of the pad and the posture of the pad.

2. The method for positioning the anchor drill drum repairing tooth seat based on three-dimensional point cloud data processing according to claim 1 is characterized in that: In the step S1, the point cloud data is obtained by obtaining the original point cloud of the drum pad surface and the nearby drum surface through binocular vision, including setting camera exposure parameters, triggering camera image acquisition and grating light emission at the same time, and obtaining the three-dimensional point cloud of the pad and its surrounding surface through phase dephasing of the acquired image.

3. The method for positioning the anchor drill drum repairing tooth seat based on three-dimensional point cloud data processing according to claim 1 is characterized in that: In the step S2, preprocessing the original point cloud refers to downsampling and outlier removal of the original point cloud. The cube size is set artificially through the moving cube method, and the center of the cube is gradually moved to traverse the entire point cloud area. Each time it moves, the points in the cube are obtained and the average is simplified to one point to achieve downsampling. The point cloud is traversed again. If the distance from the traversed point to other points is large, the current point is considered to be noise and is removed.

4. The method for positioning the anchor drill drum repairing tooth seat based on three-dimensional point cloud data processing according to claim 1 is characterized in that: In the step S3, the maximum connected domain is found by clustering the point cloud based on Euclidean distance, dividing the point cloud into several clusters, and selecting the cluster with the largest number of points as the point cloud for the next step of analysis.

5. The method for positioning the anchor drill drum repairing tooth seat based on three-dimensional point cloud data processing according to claim 1 is characterized in that: In the step S4, the 2D center is determined by setting the Z coordinates of all points in the point cloud to 0, calculating the bounding box to obtain the point cloud center, and constructing a KDtree to search for the nearest point, looking for a point with the same ID as the 2D center of the point cloud.

6. The method for positioning the anchor drill drum repairing tooth seat based on three-dimensional point cloud data processing according to claim 1 is characterized in that: In the step S6, curvature segmentation refers to calculating the curvature of the point cloud, removing points with curvature values ​​higher than a threshold, constructing a KDTree for the point cloud, calculating the normal vector of each point, traversing the point cloud, selecting the traversed point and several points closest to it, and calculating the angle between the normal vector of the traversed point and the adjacent points.

7. The method for positioning the anchor drill drum repairing tooth seat based on three-dimensional point cloud data processing according to claim 1 is characterized in that: In the step S7, the pad surface point cloud is obtained by clustering the point cloud based on Euclidean distance, calculating the point cloud centroid for each cluster, and selecting the cluster closest to the center of the original point cloud, which is the pad surface point cloud.

8. The method for positioning the anchor drill drum repairing tooth seat based on three-dimensional point cloud data processing according to claim 1 is characterized in that: In the step S8, the surface pose of the pad is calculated by calculating the OBB bounding box of the point cloud, using the upper surface center point and posture of the OBB bounding box itself to represent the surface pose of the pad, so as to quickly calculate the irregular surface pose.

9. The method for positioning the anchor drill drum repairing tooth seat based on three-dimensional point cloud data processing according to claim 1 is characterized in that: In the step S9, calculating the conversion relationship between the posture of the gear seat fitting the pad surface and the posture of the pad means converting the identified posture of the pad in the robot coordinate system into a matrix expression, inverting it through LU decomposition, and multiplying it with the matrix expression of the robot posture when the gear seat is accurately fitted, so as to obtain the conversion relationship between the posture of the gear seat fitting the pad surface and the posture of the pad.

10. A gear seat positioning device for repairing a bolter drum based on three-dimensional point cloud data processing, characterized in that: The method for positioning the tooth seat of the anchor mining machine drum repair based on three-dimensional point cloud data processing according to any one of claims 1 to 9 comprises a cutting drum body (5) and a multi-axis motion mechanism (6), the cutting drum body (5) is provided with a cushion block (4) on the outside, the cutting drum body (5) is provided with a tooth seat (3) on the outside, the tooth seat (3) is provided with a tooth seat socket (1) on the inside, the surface of the tooth seat (3) is designed with a tooth seat arc surface (2), and the multi-axis motion mechanism (6) is provided with a binocular vision sensor (7) on the outside.

Images