Path planning and control method for a curved surface constant force polishing robot

Abstract

The present invention discloses a path planning and control method for a curved surface constant-force polishing robot, comprising the following steps: dividing different areas according to the curvature of the polishing surface, and using grinding wheels of different diameters for different curvatures; determining the posture of the polishing robot at each point on the curved surface; determining the running trajectory of the robot in each area; at each posture point, making the polishing force of each point basically the same through constant force control; after determining the location of obvious defects by comparing the actual scanning model with the three-dimensional model, planning the polishing trajectory for the defective position; after completing the polishing area within the curvature range, replacing the end polishing wheel and polishing the area within the next curvature range; finally, driving the polishing robot under constant force control to complete the overall curved surface polishing.

Description

Technical Field

[0001] The present invention relates to the technical field of mechanical processing equipment, and in particular to a path planning and control method of a curved surface constant force grinding robot. Background Art

[0002] Large curved surfaces are usually obtained by casting, but the surface quality of the curved surfaces obtained by casting is poor, and secondary processing is needed to improve the surface quality, such as grinding and polishing. At present, most companies still use manual grinding, which is inefficient and has unstable grinding quality. With the rapid development of robot technology, robot grinding technology has been more and more widely used. Therefore, it is very necessary to carry out path planning and control of curved surface grinding based on industrial robots to improve grinding efficiency and grinding quality. Summary of the invention

[0003] In order to solve the problems existing in the prior art, the present invention provides a path planning and control method for a curved surface constant force polishing robot with reasonable trajectory planning and control process.

[0004] The technical solution of the present invention is as follows:

[0005] A path planning and control method for a curved surface constant force polishing robot comprises the following steps:

[0006] 1) Grinding area division: Through curvature analysis, the grinding area is divided according to the size of the curvature, and each grinding area corresponds to a model of grinding wheel;

[0007] 2) Point cloud processing: obtain point cloud data by uniformly sampling the three-dimensional model of the surface and perform noise reduction processing;

[0008] 3) Posture calculation: perform posture calculation on the point cloud data to improve the posture information of the path points;

[0009] 4) Trajectory planning: According to the grinding wheels selected in different curvature areas, the spacing between two rows of path points is selected, and the interval between adjacent path points is 1 mm. The path points in the same row are connected to obtain the robot running trajectory of the row;

[0010] 5) Point cloud comparison: Use a 3D scanner to scan the actual surface, and use the distance between the actual point cloud and the 3D model as the criterion to determine whether there are obvious defects. If there are, the defective area will be processed first, and then run according to the planned trajectory after processing;

[0011] 6) Constant force control for grinding: When the robot runs along the planned trajectory, in order to ensure that the same normal force is required for grinding, a force sensor is added to the end of the robot, and position-based impedance control is used to adjust the robot's running posture in real time to achieve constant force grinding along the curved surface.

[0012] 7) After completing the grinding area of ​​this curvature level, replace the end grinding wheel and grind the area of ​​the next curvature range.

[0013] Furthermore, the polishing area division step specifically includes the following steps:

[0014] 1.1) 3D model preprocessing, extracting the outer surface to be processed of the curved 3D model;

[0015] 1.2) Curvature analysis: Analyze the curvature of the outer surface to be processed and divide it into three grades: large, medium and small according to the size of the curvature radius;

[0016] 1.3) Area division: After dividing the grades according to the curvature radius, the adjacent areas of the same curvature grade are linked. The connection rule is to connect the large ones first and the small ones later, leaving the small curvature to the end, which is the same as the processing order.

[0017] Furthermore, the point cloud processing steps are specifically as follows:

[0018] 2.1) Discretize the surface to be processed;

[0019] 2.2) De-noising the discretized point cloud;

[0020] 2.3) Uniformly select points from the denoised point cloud.

[0021] Furthermore, the posture calculation step is specifically as follows:

[0022] 3.1) Normal calculation: the Canny operator is used to calculate the normal of the pose point, with the direction of contact with the robot as the positive direction;

[0023] 3.2) X-direction calculation: the line connecting two adjacent points along the processing direction is taken as the X-direction, with the processing direction as the positive direction;

[0024] 3.3) Y-direction calculation: After determining the X-direction and the Z-direction, use the right-hand rule to determine the Y-direction.

[0025] Furthermore, the trajectory planning steps are as follows:

[0026] 4.1) Spacing determination: the spacing between adjacent paths is set according to the partition, that is, the grinding wheel diameter used in the partition; on the same path, the point cloud interval is set to 1mm;

[0027] 4.2) Comprehensive path planning for the same-level partitions. Comprehensive path planning is carried out for the areas of the same level, that is, unified path planning is carried out for the areas of the same curvature level to determine the processing sequence, as well as the processing starting and end points of each area, so that the distance between the end point of one area and the starting point of the next area is less than 10cm.

[0028] Furthermore, the point cloud comparison step is specifically as follows:

[0029] 5.1) Scanning: Scan the processed surface to obtain point cloud data;

[0030] 5.2) Comparison: compare the scanned point cloud with the point cloud data of the 3D model, and use the distance between the pose points as the critical criterion;

[0031] 5.3) Processing: For defective areas, pre-processing trajectory planning is performed. For example, for burr parts, teaching pre-processing is used to make the defect less than 2mm and within the range of constant force grinding.

[0032] Furthermore, the grinding constant force control steps are as follows:

[0033] 6.1) Force acquisition: the force sensor installed at the end of the robot collects the grinding force information in real time;

[0034] 6.2) Information calculation: input the force information into the position-based impedance control model to calculate the position offset;

[0035] The beneficial effects of the present invention are as follows: it solves the problem of balancing efficiency and quality in path planning in large-scale curved surface grinding; it ensures grinding efficiency by adopting different grinding sheets and path planning for different curvatures, and ensures grinding quality by combining constant force control and posture calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a flowchart of the present invention;

[0037] Figure 2 A flowchart of the curvature analysis of the present invention;

[0038] Figure 3 A flowchart of the posture calculation of the present invention;

[0039] Figure 4 It is a flowchart of the scanning comparison of the present invention. DETAILED DESCRIPTION

[0040] The implementation process of the present invention will be described in more detail below with reference to the accompanying drawings. Although the process of the present disclosure is shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the implementation described here. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0041] refer to Figure 1 , which is the overall flow chart of the present invention.

[0042] A path planning and control method for a curved surface constant force polishing robot comprises the following steps:

[0043] Step A1, extracting a curved surface three-dimensional model;

[0044] Specifically, in the preprocessing step of the overall three-dimensional model of the workpiece, the outer surface to be processed of the large curved three-dimensional model is extracted;

[0045] Step A2, curvature analysis, analyzes the curvature of the outer surface to be processed, and divides it into three grades: large, medium, and small according to the size of the curvature radius;

[0046] Specifically, see the detailed steps. Figure 2 ; After the surface model is imported, the curvature of the model is calculated. After the calculation, enter the judgment module B3 and divide the area according to the curvature. For those with a curvature radius greater than 100, it is divided into a large curvature area. For those less than 100, enter the judgment module B4. For those less than 50, it is divided into a small curvature area. For those greater than 50 and less than 100, it is divided into a medium curvature area. After the division, enter B5 to integrate the discrete areas and complete the curvature and point cloud analysis steps.

[0047] Step A3, point cloud processing: discretize the model to obtain a point cloud, and process the point cloud.

[0048] Specifically, the outer surface to be processed of the curved three-dimensional model is discretized into a point cloud, and the discretized point cloud is subjected to noise reduction processing.

[0049] Step A4, posture calculation, performs posture calculation on the point cloud to determine the posture of the robot when it moves to the point.

[0050] Specifically, Figure 3 As shown, C1 uses the grinding wheel model according to the curvature partition; C2 determines the track spacing; C3 simplifies the existing point cloud data, C4 obtains the points on the track; C5 uses the Canny operator to calculate the points on the track; C6 uses the Canny calculation result to get the normal of the path point with the direction away from the robot end as the direction; C7 is the X-direction calculation, with the line connecting two adjacent points along the processing direction as the X-direction, and the processing direction as the positive; C8 is the Y-vector calculation, after the X-direction and the Z-direction are calculated, the Y-direction is determined by the right-hand rule; C9 summarizes the above three directions to get the posture of the path point.

[0051] Step A5, trajectory planning step, connects the pose points in the area in series and connects the areas within the same curvature range.

[0052] Specifically, it includes a spacing determination step, in which the spacing between adjacent paths is set according to the partition (i.e., the diameter of the grinding wheel used in the partition); on the same path, the point cloud interval is set to 1 mm; for the same-level partition comprehensive path planning, the areas of the same level are subjected to comprehensive path planning to determine the order of processing, as well as the starting and ending points of processing for each area, so that the distance between the end point of one area and the starting point of the next area is as small as possible.

[0053] Step A6, the robot running trajectory is compared with the 3D model according to the actual situation to determine whether there are obvious casting defects on the processed surface, and special treatment is performed according to the defects.

[0054] Specifically, D1 scans the actual situation of the processed surface to obtain point cloud data. D2 compares the scanned point cloud with the point cloud data of the 3D model, and uses the distance between the pose points as the critical standard. D3 is a processing step to plan the pre-processing trajectory for the defective area. D4 runs according to the original trajectory.

[0055] Step A7, robot constant force control, using position-based impedance control.

[0056] Specifically, the grinding force information is collected in real time through the force sensor installed at the end of the robot; the force information is input into the position-based impedance control model to calculate the position offset; the position offset is input into the robot to control the robot to run to the corresponding corrected posture to achieve constant force grinding of the curved surface.

[0057] Step A8, actual running trajectory after constant force control.

[0058] Step A9, after grinding an area with one curvature level, replace the knife to grind an area with the next curvature level.

[0059] Specifically, after the robot finishes grinding the large curvature area, it replaces the grinding wheel with a smaller model to grind the medium curvature area, and finally replaces the grinding wheel again to grind the small curvature area to complete the grinding of the entire curved surface.

[0060] The above embodiments are only preferred embodiments of the present invention and are not limitations of the technical solutions of the present invention. Any technical solution that can be implemented on the basis of the above embodiments without creative work should be deemed to fall within the scope of protection of the patent of the present invention.

Claims

1. A path planning and control method for a curved surface constant force polishing robot, characterized in that: The following steps are involved: 1) Grinding area division: Through curvature analysis, the grinding area is divided according to the size of the curvature, and each grinding area corresponds to a model of grinding wheel; The polishing area division step specifically includes the following steps: 1.1) 3D model preprocessing, extracting the outer surface to be processed of the curved 3D model; 1.2) Curvature analysis: Analyze the curvature of the outer surface to be processed and divide it into three grades: large, medium and small according to the size of the curvature radius; 1.3) Area division: After dividing the grades according to the curvature radius, connect the adjacent areas of the same curvature grade. The connection rule is to connect the large ones first and the small ones later, leaving the small curvature to the end, which is the same as the processing order; 2) Point cloud processing: obtain point cloud data by uniformly sampling the three-dimensional model of the surface and perform noise reduction processing; The point cloud processing steps are as follows: 2.1) Discretize the surface to be processed; 2.2) De-noising the discretized point cloud; 2.3) Evenly select points from the denoised point cloud; 3) Posture calculation: perform posture calculation on the point cloud data to improve the posture information of the path points; The posture calculation steps are as follows: 3.1) Normal calculation: the Canny operator is used to calculate the normal of the pose point, with the direction of contact with the robot as the positive direction; 3.2) X-direction calculation: the line connecting two adjacent points along the processing direction is taken as the X-direction, with the processing direction as the positive direction; 3.3) Y-direction calculation: After determining the X-direction and the Z-direction, use the right-hand rule to determine the Y-direction; 4) Trajectory planning: According to the grinding wheels selected in different curvature areas, the spacing between two rows of path points is selected, and the interval between adjacent path points is 1 mm. The path points in the same row are connected to obtain the robot running trajectory of the row; The trajectory planning steps are as follows: 4.1) Spacing determination: the spacing between adjacent paths is set according to the partition, that is, the grinding wheel diameter used in the partition; on the same path, the point cloud interval is set to 1mm; 4.2) Comprehensive path planning for the same-grade partitions: conduct comprehensive path planning for the same-grade areas, that is, conduct unified path planning for the same-grade areas, determine the processing sequence, and the processing start and end points of each area, so that the distance between the end point of one area and the start point of the next area is less than 10cm; 5) Point cloud comparison: Use a 3D scanner to scan the actual surface, and use the distance between the actual point cloud and the 3D model as the criterion to determine whether there are obvious defects. If there are, the defective area will be processed first, and then run according to the planned trajectory after processing; 6) Grinding constant force control: When the robot runs along the planned trajectory, in order to ensure that the same material is removed, the normal force of grinding must be the same. A force sensor is added to the end of the robot, and position-based impedance control is used to adjust the robot's running posture in real time to achieve constant force grinding along the curved surface. The grinding constant force control steps are as follows: 6.1) Force acquisition: the force sensor installed at the end of the robot collects the grinding force information in real time; 6.2) Information calculation: input the force information into the position-based impedance control model to calculate the position offset; 6.3) Control the robot, input the position offset to the robot, control the robot to run to the corrected corresponding posture, and realize constant force grinding of the curved surface; 7) After completing the grinding of one curvature level area, replace the end grinding wheel and grind the area of ​​the next curvature range.

2. The path planning and control method of a curved surface constant force polishing robot according to claim 1, characterized in that: The point cloud comparison steps are as follows: 5.1) Scanning: Scan the processed surface to obtain point cloud data; 5.2) Comparison: compare the scanned point cloud with the point cloud data of the 3D model, and use the distance between the pose points as the critical criterion; 5.3) Processing: Pre-processing trajectory planning for defective areas; For burr parts, use teaching pre-treatment to make the defect less than 2mm and within the range of constant force grinding.

Images