Tire pattern block double-process precision machining device and machining control method
By using a binocular vision-based 3D surface reconstruction system, combined with offline calibration and teaching, the problems of low precision and poor consistency in the precision machining of tire tread blocks were solved, achieving high-precision chamfering and milling, and ensuring consistent machining quality.
Patent Information
- Application Number
- CN202511227155.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-29
- Publication Date
- 2025-11-21
AI Technical Summary
In the existing technology, the precision machining of tire tread blocks relies on manual teaching, which results in low precision, poor consistency, and low efficiency. Furthermore, binocular vision 3D reconstruction is easily affected by lighting and weak texture interference, making it difficult to achieve high-precision integrated machining.
A 3D surface reconstruction system based on binocular vision is adopted, which combines offline calibration and teaching, pattern block surface reconstruction and hole and groove positioning, tool setting and vision-guided machining. By using a customized arc light source in conjunction with a binocular camera, the imaging quality is improved, and the accuracy of the vision-guided robot is improved through multiple teachings, ultimately achieving chamfering and milling of pattern blocks with a diameter of more than 0.2mm.
It achieves high-precision and stable chamfering and milling of patterned blocks, ensuring the consistency of machining quality of all back holes, overcoming the limitation of insufficient consistency in depth estimation by binocular vision, and improving the machining accuracy of the robot in a wide range of multi-angle conditions.
Smart Images
Figure CN120985741A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of machine vision technology, and in particular to a dual-process precision machining device and control method for tire tread blocks based on binocular vision three-dimensional surface reconstruction. Background Technology
[0002] Currently, precision machining of tire tread blocks largely relies on manual teaching, resulting in low accuracy, poor consistency, and low efficiency, making it difficult to meet processing requirements. The severe reflectivity of the tire tread block surface significantly impacts the imaging performance of laser line scanning cameras and structured light 3D cameras. While binocular vision can provide 3D information, it is susceptible to interference from lighting, weak textures, and repetitive textures, leading to high noise and low accuracy in reconstructed point clouds, making it unsuitable for direct precision machining. In existing technical solutions, 3D reconstruction and positioning, and CNC machining are often two independent processes. The reconstructed data often requires complex offline processing and data conversion to generate a machining path, failing to achieve integrated closed-loop control of "measurement-reconstruction-machining," making it difficult to meet the real-time requirements of production lines, and resulting in low efficiency.
[0003] This invention proposes a dual-process precision machining device and control method for tire tread blocks based on binocular vision 3D surface reconstruction. This is an integrated system that combines high-precision 3D reconstruction and real-time machining control. Through three stages—offline calibration and teaching, tread block surface reconstruction and hole / groove positioning, and tool setting and vision-guided machining—it systematically overcomes the limitations of insufficient consistency in depth estimation caused by binocular vision on precision machining results. Ultimately, it completes the chamfering and milling of tread blocks with a precision requirement of over 0.2mm, while ensuring the consistency of machining quality for all back holes. Summary of the Invention
[0004] To address the shortcomings of existing technologies, this invention provides a dual-process precision machining device and machining control method for tire tread blocks. Through three stages—offline calibration and teaching, tread block surface reconstruction and hole and groove positioning, and tool setting and vision-guided machining—the device ultimately completes the chamfering and milling of tread blocks with a precision requirement of 0.2 mm or higher, while ensuring the consistency of machining quality for all back holes.
[0005] The objective of this invention is achieved as follows: a dual-process precision machining device for tire tread blocks based on binocular vision three-dimensional surface reconstruction, comprising:
[0006] Robots are used to perform chamfering and milling operations;
[0007] Vision control system, including industrial computer and vision host computer software;
[0008] A binocular camera, mounted on a fixed bracket, is used to achieve high-precision reconstruction of the patterned block surface and high-precision positioning of the back hole;
[0009] The light source assembly includes a custom-designed curved light source, a custom-designed bracket, and a light source controller. The light source assembly includes a custom-designed curved surface light source installed on the top of the bracket for illumination and installed on the side of the bracket for supplemental lighting. The rotary worktable includes a motor-driven circular rotary platform and a patterned block clamping mechanism.
[0010] A rotary table is used to rotate tire tread blocks placed in a designated position;
[0011] Machining apparatus, including an electric spindle and its controller;
[0012] The control panel includes a custom-designed electrical cabinet and a touchscreen mounted on the surface of the cabinet.
[0013] The tool setting device is installed on the outside of the turntable to compensate for tool changing errors.
[0014] A dual-process precision machining control method for tire tread blocks includes three stages: offline calibration and teaching, tread block surface reconstruction and slot positioning, and tool setting and vision-guided machining.
[0015] Offline calibration and teaching phase:
[0016] Step S1) Calibrate the binocular camera parameters, hand-eye relationship, relative pose between the turntable rotation center coordinate system and the robot coordinate system, and relative pose between the turntable surface coordinate system and the left camera.
[0017] Step S2) Evenly distribute and fix multiple calibration plates on the surface of the patterned block, take pictures of the calibration plates, teach the robot pose when the pendant is aligned with the center of each calibration plate, calculate and record the mapping relationship between the calibration plate pose and the robot alignment pose.
[0018] Pattern block surface reconstruction and slot positioning stage:
[0019] Step S3) After taking pictures of the patterned block, the binocular image pairs are first corrected. Then, the distance between the camera and the turntable surface is converted into the disparity value in the binocular vision system. The disparity value of each pixel is calculated through the multi-grid optimized stereo matching algorithm. Finally, the surface point cloud is generated based on the disparity map.
[0020] Step S4) Fit the topmost generatrix of the surface based on the disparity map, and reconstruct the surface point cloud by combining the radius of the pattern block;
[0021] Step S5) Locate the position of the air groove in the patterned block, combine the width and depth information of the air groove to generate the point cloud data of the air groove, and fuse the air groove point cloud in the curved surface point cloud.
[0022] Step S6) Locate the back hole of the patterned block, take the point cloud around it to fit the center coordinates of the hole, and calculate the position of the nearest milling groove;
[0023] Step S7) Calculate the points required for robot-guided processing based on the mapping relationship saved during teaching and send them to the robot;
[0024] Tool setting and vision-guided machining stage:
[0025] Step S8) During operation, the tool is first set using a tool setter. The robot saves the tool setting results, and the visual point position is offset according to the tool setting data and the touch screen parameters. Finally, chamfering and milling are performed.
[0026] Furthermore, step S1 specifically includes:
[0027] First, calibrate the intrinsic parameters of the left and right cameras, and then calibrate the relative pose relationship between the left and right cameras. The relative pose relationship between the left camera and the robot's base coordinate system is calibrated. Then, the relative pose relationship between the rotary table rotation center coordinate system and the robot base coordinate system is calibrated. The specific steps are as follows: First, calibrate the tool coordinate system using the robot's internal calibration method. The origin of the tool coordinate system is at the tool tip position. Fix the calibration plate on the turntable plane and rotate the turntable. After each rotation, control the robot via the teach pendant to align the tool tip with the center of the calibration plate in the same posture, and record the tool coordinate system pose. After rotating 360°, obtain N poses, which are the coordinates of the calibration plate center in the robot's base coordinate system during N rotations. Fit the N coordinates to a plane, take the plane normal as the direction of the turntable rotation axis, and take the center of the rotation axis as the center of the fitted plane. Finally, place the calibration plate at the pattern block clamping position on the turntable surface, take a picture, and calibrate the relative pose relationship between the turntable surface coordinate system and the left camera. .
[0028] This step calibrates the system, providing a foundation for subsequent 3D reconstruction and coordinate transformation.
[0029] Furthermore, step S2 specifically involves:
[0030] Multiple calibration plates are evenly distributed and attached to the surface of the patterned block, with the plane of the calibration plate tangent to the surface of the patterned block. The calibration plates are photographed and positioned. The turntable is rotated 180° so that the patterned block reaches the processing position. Based on the calibration data in step S1, the pose of the calibration plate in the robot base coordinate system is calculated and recorded. The robot is controlled so that the cutting head is aligned with the center of the calibration plate in sequence. The pose of the robot flange coordinate system relative to the robot base coordinate system is recorded, and the mapping relationship between the pose of the calibration plate and the robot alignment pose is calculated.
[0031] By teaching and calculating multiple mapping relationships, the teaching data closest to the positioning point can be selected to calculate the alignment pose during automated operation, effectively improving the accuracy and stability of the robot vision system.
[0032] Furthermore, step S3 specifically includes:
[0033] After acquiring images of the patterned blocks, stereo correction is first performed on the acquired binocular image pairs. Then, based on the camera parameters calibrated in step S1, the physical distance between the camera optical center and the turntable surface is converted into the corresponding theoretical disparity value. Referring to this disparity range, the disparity of the corrected stereo image pairs is calculated based on the variational method and a multi-grid solver. The optimal disparity is determined by minimizing the energy function. At the same time, the multi-grid method is used, combined with coarse-to-fine image pyramid initialization, fixed-point iterative approximation, and efficient V / W loop solution of linear equations. Finally, a dense, high-precision disparity map that can cover textureless areas is output. Then, point cloud data of the patterned block surface is generated by combining the corrected image pairs and correction parameters.
[0034] This method effectively improves the estimation accuracy and robustness of depth information in binocular stereo vision systems, and in particular solves the problem of binocular 3D measurement on highly reflective and textured surfaces, providing a reliable data foundation for subsequent 3D reconstruction and accurate measurement.
[0035] Furthermore, step S4 specifically involves:
[0036] The disparity map obtained in step S3 is divided into multiple rectangular regions. The point cloud in each region is fitted into a three-dimensional curve, resulting in multiple three-dimensional curves. These curves are then combined with the known geometric constraint of the pattern block radius. Multiple 3D points close to the upper edge generatrix were obtained. Projection and fitting algorithms were used to smooth and optimize the generatrix. Based on the generatrix direction, a coordinate system for the patterned block surface was established, with the origin at the midpoint of the fitted generatrix and the X-axis parallel to the generatrix direction. Determine other degrees of freedom. Based on the coordinate system of the patterned block surface and the known radius information, generate multiple three-dimensional curves perpendicular to the generatrix, thereby obtaining a high-precision, smooth three-dimensional curved surface point cloud model.
[0037] This method effectively overcomes the noise and uncertainty caused by simple binocular vision measurement, and in particular solves the problem of image edge through-hole deformation and binocular matching difficulties in this scenario. It significantly improves the accuracy and reliability of surface reconstruction and ensures the consistency of subsequent dual-process precision machining quality.
[0038] Furthermore, step S5 specifically involves:
[0039] The approximate area of the air groove is obtained by using a deep learning semantic segmentation algorithm. The boundary of the air groove is accurately located by combining morphological analysis and edge extraction methods. By combining the design width and depth parameters of the air groove, air groove point cloud data with cross-sectional shape is generated along the generatrix direction. The data is then seamlessly integrated into the curved point cloud using a non-rigid registration method. The final fitted pattern block has air groove features on its surface, which is close to the surface of the real pattern block.
[0040] Furthermore, step S6 specifically includes:
[0041] The deep learning target detection algorithm is used to locate the through hole on the back of the patterned block, which effectively reduces the negative impact of the imaging angle and distortion of the through hole on the stability of the positioning. After obtaining the two-dimensional coordinates of the through hole, the curved point cloud is projected onto the two-dimensional image, the distance between the through hole and the projection point of the point cloud is calculated, the point cloud near the through hole is selected, and the accurate chamfer pose is obtained by fitting the plane according to the point cloud. At the same time, the position of the nearest air groove is calculated.
[0042] By fitting the plane with the point cloud near the through hole, the positioning accuracy of the through hole at the edge of the air groove is effectively improved.
[0043] Furthermore, step S7 specifically includes:
[0044] Calculate the distance between the through hole and the center of the calibration plate during teaching, select the teaching data with the closest distance, find the relative pose of the flange and the teaching point during the corresponding precise alignment, and calculate the precise pose of the tool head aligned with the through hole and the corresponding air groove.
[0045] Furthermore, step S8 specifically includes:
[0046] During tool setting, the tool setter communicates with the robot. When the robot receives the tool setter's positioning signal, it calculates and records the difference between the previous tool setting data and the previous data. Based on this difference and the chamfer and milling groove depth set by the operator on the touch screen, the visual point is shifted along the flange's Z-axis. Chamfering and milling are then performed based on the shifted point.
[0047] Compared with existing technologies, the advantages of this invention are as follows: This invention, through the combination of a customized arc-shaped light source and binocular vision, solves the problem of the impact of highly reflective materials on imaging quality of tire tread blocks, resulting in clearer and more stable imaging compared to other 3D systems. The curved surface 3D reconstruction method proposed in this invention systematically overcomes the limitations imposed on precision machining effects by insufficient consistency in depth estimation in binocular vision. Simultaneously, by employing a multiple-teaching method, it improves the guidance accuracy of the vision-guided robot when performing large-scale, multi-angle alignment, ultimately completing the chamfering and milling of tread blocks with an accuracy requirement of over 0.2mm, while ensuring the consistency of machining quality for all back holes. Previously, the final alignment accuracy of robot binocular 3D vision was generally below 0.5mm, and could only guarantee alignment accuracy within a small working space, with inconsistent measurement depths and poor binocular matching stability. This device and method can be applied to various high-reflectivity curved surface machining scenarios. By changing the end effector, it can achieve tasks such as large-scale, multi-angle curved surface machining with an accuracy of 0.2mm. Attached Figure Description
[0048] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.
[0049] Figure 1 This is a schematic diagram of the dual-process precision machining device for tire tread blocks in this invention.
[0050] Figure 2 This is a flowchart of the precision machining control method for dual-process tire tread blocks in this invention.
[0051] Figure 3 This is a schematic diagram of the calibration board image positioning during the teaching process in this invention.
[0052] Figure 4 This is a physical image of the tire tread block in this invention.
[0053] Figure 5 This is a surface point cloud map generated based on a binocular image in this invention.
[0054] Figure 6 This is a schematic diagram of the busbar fitting process in this invention.
[0055] Figure 7 This is a schematic diagram of the edge positioning of the air groove in this invention.
[0056] Figure 8 This is the complete surface point cloud reconstructed in this invention.
[0057] Figure 9 This is the interface for displaying the positioning results of the host computer software in this invention.
[0058] Figure 10 This is the touchscreen parameter setting interface in this invention.
[0059] Figure 11 This is an alignment effect diagram of the processing in this invention.
[0060] The components include: 1. Robot; 2. Vision control system; 3. Binocular camera; 4. Fixed bracket; 5. Light source assembly; 6. Rotary worktable; 7. Processing device; 8. Operating table; and 9. Tool setting device. Detailed Implementation
[0061] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0062] Example 1
[0063] like Figure 1 The device shown is a dual-process precision machining apparatus for tire tread blocks based on binocular vision 3D surface reconstruction. It includes a robot 1 for chamfering and milling operations; a vision control system 2, including an industrial computer and vision host computer software; a binocular camera 3, mounted on a fixed bracket 4, for high-precision reconstruction of the tread block surface and high-precision positioning of the back hole; a light source assembly 5, including a custom-designed arc light source, a custom bracket, and a light source controller; a rotary table 6 for rotating the tire tread block placed in a designated position; a machining device 7, including an electric spindle and its controller; an operating table 8, including a custom-designed electrical cabinet and a touch screen mounted on the cabinet surface; and a tool setting device 9, mounted on the outside of the rotary table, for compensating for tool changing errors.
[0064] Example 2
[0065] like Figure 2 The method for controlling the precision machining of tire tread blocks using a dual-process approach employs the teaching device described in Example 1 and includes three stages: offline calibration and teaching, tread block surface reconstruction and slot positioning, and tool setting and vision-guided machining.
[0066] Offline calibration and teaching phase:
[0067] Step S1: Calibrate the binocular camera parameters, hand-eye relationship, relative pose between the turntable rotation center coordinate system and the robot coordinate system, and relative pose between the turntable surface coordinate system and the left camera;
[0068] Step S2: Evenly distribute and fix the fifteen calibration plates on the surface of the patterned block, and take pictures of the calibration plates (e.g., ...). Figure 3 The robot pose when the teaching pendant is aligned with the center of each calibration board is calculated and recorded.
[0069] Pattern block surface reconstruction and slot positioning stage:
[0070] Step S3: Apply the patterned blocks (e.g.) Figure 4 After taking the picture, the binocular image pairs are first calibrated. Then, the distance between the camera and the turntable surface is converted into disparity values in the binocular vision system. A multi-grid optimized stereo matching algorithm is used to calculate the disparity value of each pixel. Finally, a surface point cloud (e.g., ...) is generated based on the disparity map. Figure 5 );
[0071] Step S4: Fit the topmost generatrix of the surface based on the disparity map (e.g., Figure 6 ), and reconstruct the surface point cloud by combining the radius of the patterned blocks;
[0072] Step S5: Locate the air groove position of the patterned block (e.g.) Figure 7 By combining the width and depth information of the air vents, point cloud data of the air vents is generated, and the air vent point cloud is fused into the curved surface point cloud (e.g., Figure 8 );
[0073] Step S6: Locate the back hole of the patterned block, obtain the center coordinates of the hole by fitting the point cloud around it, and calculate the nearest milling groove position (e.g., Figure 9 ).
[0074] Step S7: Calculate the points required for robot-guided processing based on the mapping relationship saved during teaching and send them to the robot;
[0075] Tool setting and vision-guided machining stage:
[0076] Step S8: During operation, the tool is first set using a tool setter. The robot saves the tool setting results and compares the tool setting data with the parameters on the touchscreen (e.g., ...). Figure 10 The visual point is shifted, and finally chamfering and milling are performed (e.g.) Figure 11 ).
[0077] Step S1 is as follows:
[0078] First, calibrate the intrinsic parameters of the left and right cameras, and then calibrate the relative pose relationship between the left and right cameras. The relative pose relationship between the left camera and the robot's base coordinate system is calibrated. Then, the relative pose relationship between the rotary table rotation center coordinate system and the robot coordinate system is calibrated. The specific steps are as follows: First, calibrate the tool coordinate system using the robot's internal calibration method. The origin of the tool coordinate system is at the cutter head (tip). Fix the calibration plate on the turntable plane and rotate the turntable in 30° increments. After each rotation, control the robot via the teach pendant to align the cutter head with the center of the calibration plate in the same posture. Record the tool coordinate system pose. After rotating 360°, obtain 12 poses, which are the coordinates of the calibration plate center in the robot's base coordinate system during the 12 rotations. Fit the 12 coordinates to a plane, taking the plane normal as the direction of the turntable rotation axis, and taking the center of the rotation axis as the center of the fitted plane. Finally, place the calibration plate at the pattern block clamping position on the turntable surface, take a picture, and calibrate the relative pose relationship between the turntable surface coordinate system and the left camera. .
[0079] .
[0080] Step S2 is as follows:
[0081] Fifteen calibration plates are evenly distributed and attached to the surface of the patterned block, with the plane of the calibration plate tangent to the surface of the patterned block. The 15 calibration plates are photographed and positioned. The turntable is rotated 180° so that the patterned block reaches the processing position. Based on the calibration data in step S1, the poses of the 15 calibration plates in the robot base coordinate system are calculated and recorded. The robot is controlled so that the cutting head is aligned with the center of the 15 calibration plates in sequence. The poses of the 15 robot flange coordinate systems relative to the robot base coordinate system are recorded, and the mapping relationship between the 15 sets of calibration plate poses and the robot alignment poses is calculated.
[0082] By teaching and calculating multiple mapping relationships, the teaching data closest to the positioning point can be selected to calculate the alignment pose during automated operation, effectively improving the accuracy and stability of the robot vision system.
[0083] Step S3 is as follows:
[0084] After acquiring images of the patterned blocks, stereo correction is first performed on the acquired binocular image pairs to ensure strict epipolar alignment between the two images, satisfying the geometric constraints of stereo matching. Subsequently, based on the camera parameters calibrated in step S1, the physical distance (manually measured) between the camera's optical center and the turntable surface is converted into the corresponding theoretical disparity value. Referring to this disparity range, the disparity of the corrected stereo image pairs is calculated using a variational method and a multigrid solver. The core of this calculation is an energy function constructed using the variational method, which includes a data term (constraining the grayscale and gradient constancy of corresponding points in the image) and a smoothing term (ensuring piecewise smoothness of the disparity field). The optimal disparity is determined by minimizing the energy function. Simultaneously, a multigrid method is employed, combining coarse-to-fine image pyramid initialization, fixed-point iterative approximation, and efficient V / W loop solving of linear equations. The final output is a dense, high-precision disparity map that covers textureless areas. This map is then combined with the corrected image pairs and correction parameters to generate point cloud data of the patterned block surface.
[0085] This method effectively improves the estimation accuracy and robustness of depth information in binocular stereo vision systems, and in particular solves the problem of binocular 3D measurement on highly reflective and textured surfaces, providing a reliable data foundation for subsequent 3D reconstruction and accurate measurement.
[0086] Step S4 is as follows:
[0087] The disparity map obtained in step S3 is divided into 5 rectangular regions. The point cloud in each region is fitted into a 3D curve, resulting in 5 3D curves. These curves are then combined with the known geometric constraint of the pattern block radius. Five 3D points close to the upper edge generatrix were obtained. The generatrix was smoothed and optimized using projection and fitting algorithms. A coordinate system for the patterned block surface was established based on the generatrix direction, with the origin at the midpoint of the fitted generatrix and the X-axis parallel to the generatrix direction. Determine other degrees of freedom. Based on the surface coordinate system of the patterned block and the known radius information, generate multiple three-dimensional curves perpendicular to the generatrix with a step size of 0.01 mm, thereby obtaining a high-precision, smooth three-dimensional surface point cloud model.
[0088] This method effectively overcomes the noise and uncertainty caused by simple binocular vision measurement, and in particular solves the problem of image edge through-hole deformation and binocular matching difficulties in this scenario. It significantly improves the accuracy and reliability of surface reconstruction and ensures the consistency of subsequent dual-process precision machining quality.
[0089] Step S5 is as follows:
[0090] The approximate area of the air groove is obtained by using a deep learning semantic segmentation algorithm. The boundary of the air groove is accurately located by combining morphological analysis and edge extraction methods. By combining the design width and depth parameters of the air groove, air groove point cloud data with cross-sectional shape is generated along the generatrix direction. The data is then seamlessly integrated into the curved point cloud using a non-rigid registration method. The final fitted pattern block has air groove features on its surface, which is close to the surface of the real pattern block.
[0091] Step S6 is as follows:
[0092] The deep learning target detection algorithm is used to locate the through hole on the back of the patterned block, which effectively reduces the negative impact of the imaging angle and distortion of the through hole on the stability of the positioning. After obtaining the two-dimensional coordinates of the through hole, the curved point cloud is projected onto the two-dimensional image, the distance between the through hole and the projection point of the point cloud is calculated, the point cloud near the through hole is selected, and the accurate chamfer pose is obtained by fitting the plane according to the point cloud. At the same time, the position of the nearest air groove is calculated.
[0093] Step S7 is as follows:
[0094] Calculate the distance between the through hole and the center of the calibration plate during 15 teaching sessions, select the teaching data closest to the distance, find the relative pose of the flange and the teaching point during the corresponding precise alignment, and calculate the precise pose of the tool head aligned with the through hole and the corresponding air groove.
[0095] Step S8 is as follows:
[0096] During tool setting, the tool setter communicates with the robot. When the robot receives the tool setter's positioning signal, it calculates and records the difference between the previous tool setting data and the previous data. Based on this difference and the chamfer and milling groove depth set by the operator on the touch screen, the visual point is shifted along the flange's Z-axis. Chamfering and milling are then performed based on the shifted point.
[0097] The above description of the embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. It should be noted that those skilled in the art can make several improvements and modifications to the present invention without departing from the principles of the present invention, and these improvements and modifications also fall within the protection scope of the claims of the present invention.
Claims
1. A precision machining device for tire tread blocks using a dual-process method, characterized in that, include: Robots are used to perform chamfering and milling operations; Vision control system, including industrial computer and vision host computer software; A binocular camera, mounted on a fixed bracket, is used to achieve high-precision reconstruction of the patterned block surface and high-precision positioning of the back hole; The light source assembly includes a custom curved light source, a custom bracket, and a light source controller. The light source assembly includes a custom curved surface light source installed on the top of the bracket for illumination and installed on the side of the bracket for supplemental lighting. The rotary table includes a motor-driven circular rotary platform and a pattern block clamping mechanism; A rotary table is used to rotate tire tread blocks placed in a designated position; Machining apparatus, including an electric spindle and its controller; The control panel includes a custom-designed electrical cabinet and a touchscreen mounted on the surface of the cabinet. The tool setting device is installed on the outside of the turntable to compensate for tool changing errors.
2. A method for controlling the precision machining of tire tread blocks using a dual-process method, employing the precision machining apparatus for tire tread blocks as described in claim 1, characterized in that... It includes three stages: offline calibration and teaching, pattern block surface reconstruction and hole and slot positioning, and tool setting and vision-guided machining. Offline calibration and teaching phase: Step S1) Calibrate the binocular camera parameters, hand-eye relationship, relative pose between the turntable rotation center coordinate system and the robot coordinate system, and relative pose between the turntable surface coordinate system and the left camera. Step S2) Evenly distribute and fix multiple calibration plates on the surface of the patterned block, take pictures of the calibration plates, teach the robot pose when the pendant is aligned with the center of each calibration plate, calculate and record the mapping relationship between the calibration plate pose and the robot alignment pose. Pattern block surface reconstruction and slot positioning stage: Step S3) After taking pictures of the patterned block, the binocular image pairs are first corrected. Then, the distance between the camera and the turntable surface is converted into the disparity value in the binocular vision system. The disparity value of each pixel is calculated through the multi-grid optimized stereo matching algorithm. Finally, the surface point cloud is generated based on the disparity map. Step S4) Fit the topmost generatrix of the surface based on the disparity map, and reconstruct the surface point cloud by combining the radius of the pattern block; Step S5) Locate the position of the air groove in the patterned block, combine the width and depth information of the air groove to generate the point cloud data of the air groove, and fuse the air groove point cloud in the curved surface point cloud. Step S6) Locate the back hole of the patterned block, take the point cloud around it to fit the center coordinates of the hole, and calculate the position of the nearest milling groove; Step S7) Calculate the points required for robot-guided processing based on the mapping relationship saved during teaching and send them to the robot; Tool setting and vision-guided machining stage: Step S8) During operation, the tool is first set using a tool setter. The robot saves the tool setting results, and the visual point position is offset according to the tool setting data and the touch screen parameters. Finally, chamfering and milling are performed.
3. The method for controlling the precision machining of tire tread blocks using a dual-process method according to claim 2, characterized in that, Step S1 is as follows: First, calibrate the intrinsic parameters of the left and right cameras, and then calibrate the relative pose relationship between the left and right cameras. The relative pose relationship between the left camera and the robot's base coordinate system is calibrated. Then, the relative pose relationship between the rotary table rotation center coordinate system and the robot base coordinate system is calibrated. The specific steps are as follows: First, calibrate the tool coordinate system using the robot's internal calibration method. The origin of the tool coordinate system is at the tool tip position. Fix the calibration plate on the turntable plane and rotate the turntable. After each rotation, control the robot via the teach pendant to align the tool tip with the center of the calibration plate in the same posture, and record the tool coordinate system pose. After rotating 360°, obtain N poses, which are the coordinates of the calibration plate center in the robot's base coordinate system during N rotations. Fit the N coordinates to a plane, take the plane normal as the direction of the turntable rotation axis, and take the center of the rotation axis as the center of the fitted plane. Finally, place the calibration plate at the pattern block clamping position on the turntable surface, take a picture, and calibrate the relative pose relationship between the turntable surface coordinate system and the left camera. .
4. The method for controlling the precision machining of tire tread blocks using a dual-process method according to claim 3, characterized in that, Step S2 is as follows: Multiple calibration plates are evenly distributed and attached to the surface of the patterned block, with the plane of the calibration plate tangent to the surface of the patterned block. The calibration plates are photographed and positioned. The turntable is rotated 180° so that the patterned block reaches the processing position. Based on the calibration data in step S1, the pose of the calibration plate in the robot base coordinate system is calculated and recorded. The robot is controlled so that the cutting head is aligned with the center of the calibration plate in sequence. The pose of the robot flange coordinate system relative to the robot base coordinate system is recorded, and the mapping relationship between the pose of the calibration plate and the robot alignment pose is calculated.
5. The method for controlling the precision machining of tire tread blocks using a dual-process method according to claim 4, characterized in that, Step S3 is as follows: After acquiring images of the patterned blocks, stereo correction is first performed on the acquired binocular image pairs. Then, based on the camera parameters calibrated in step S1, the physical distance between the camera optical center and the turntable surface is converted into the corresponding theoretical disparity value. Referring to this disparity range, the disparity of the corrected stereo image pairs is calculated based on the variational method and a multi-grid solver. The optimal disparity is determined by minimizing the energy function. At the same time, the multi-grid method is used, combined with coarse-to-fine image pyramid initialization, fixed-point iterative approximation, and efficient V / W loop solution of linear equations. Finally, a dense, high-precision disparity map that can cover textureless areas is output. Then, point cloud data of the patterned block surface is generated by combining the corrected image pairs and correction parameters.
6. The method for controlling the precision machining of tire tread blocks using a dual-process method according to claim 5, characterized in that, Step S4 is as follows: The disparity map obtained in step S3 is divided into multiple rectangular regions. The point cloud in each region is fitted into a three-dimensional curve, resulting in multiple three-dimensional curves. These curves are then combined with the known geometric constraint of the pattern block radius. Multiple 3D points close to the upper edge generatrix were obtained. Projection and fitting algorithms were used to smooth and optimize the generatrix. Based on the generatrix direction, a coordinate system for the patterned block surface was established, with the origin at the midpoint of the fitted generatrix and the X-axis parallel to the generatrix direction. Determine other degrees of freedom. Based on the coordinate system of the patterned block surface and the known radius information, generate multiple three-dimensional curves perpendicular to the generatrix, thereby obtaining a high-precision, smooth three-dimensional curved surface point cloud model.
7. The method for controlling the precision machining of tire tread blocks using a dual-process method according to claim 6, characterized in that, Step S5 is as follows: The approximate area of the air groove is obtained by using a deep learning semantic segmentation algorithm. The boundary of the air groove is accurately located by combining morphological analysis and edge extraction methods. By combining the design width and depth parameters of the air groove, air groove point cloud data with cross-sectional shape is generated along the generatrix direction. The data is then seamlessly integrated into the curved point cloud using a non-rigid registration method. The final fitted pattern block has air groove features on its surface, which is close to the surface of the real pattern block.
8. The method for controlling the precision machining of tire tread blocks using a dual-process method according to claim 7, characterized in that, Step S6 is as follows: The deep learning target detection algorithm is used to locate the through hole on the back of the patterned block, which effectively reduces the negative impact of the imaging angle and distortion of the through hole on the stability of the positioning. After obtaining the two-dimensional coordinates of the through hole, the curved point cloud is projected onto the two-dimensional image, the distance between the through hole and the projection point of the point cloud is calculated, the point cloud near the through hole is selected, and the accurate chamfer pose is obtained by fitting the plane according to the point cloud. At the same time, the position of the nearest air groove is calculated.
9. The method for controlling the precision machining of tire tread blocks using a dual-process method according to claim 8, characterized in that, Step S7 is as follows: Calculate the distance between the through hole and the center of the calibration plate during teaching, select the teaching data with the closest distance, find the relative pose of the flange and the teaching point during the corresponding precise alignment, and calculate the precise pose of the tool head aligned with the through hole and the corresponding air groove.
10. The method for controlling the precision machining of tire tread blocks using a dual-process method according to claim 9, characterized in that, Step S8 is as follows: During tool setting, the tool setter communicates with the robot. When the robot receives the tool setter's positioning signal, it calculates and records the difference between the previous tool setting data and the previous data. Based on this difference and the chamfer and milling groove depth set by the operator on the touch screen, the visual point is shifted along the flange's Z-axis. Chamfering and milling are then performed based on the shifted point.
Citation Information
Cited By
Machining control method and workpiece machining equipment
CN121787165A