Research On Robot Automatic Grinding Programming System Based On Machine Vision

With the advancement of science and technology,robots have become more widely used in the field of industrial manufacturing,and robotic grinding has become a new processing mode.However,a large portion of them is based on the use of schematic programming,which improves production efficiency and quality to a degree but requires re-showing after the replacement of parts,and lacks a certifiable standard.Using automotive brake calipers as the research object,this paper proposes a robotic automatic grinding system based on point cloud data,which is based on the combination of machine vision technology and robotic grinding technology,to improve production efficiency and processing quality by processing the obtained data.To begin,a visual recognition method suitable for this system is selected by analyzing the classification of visual recognition technologies and their corresponding application scenarios,as well as their respective advantages and disadvantages.An automatic grinding recognition system is established to obtain better point cloud and feature information of the workpiece;hand-eye calibration is performed,and the data obtained from the recognition is associated with the robot can then get information on the position of the workpiece in relation to the base coordinate system,ensuring that the subsequent grinding trajectory is accurate and reliable.The point cloud data is then pre-processed with binocular structured light technology to remove outliers,as well as downsampling and noise smoothing to reduce external noise interference and increase post-processing speed and quality.After discretizing the grinding path using the isoparametric step method for small curvature variations and the isosceles height error method for large curvature variations,the grinding path is interpolated using NURBS curve fitting,and the normal vector estimation is used to plan the robot’s grinding attitude.The generated data is then used to define the grinding path,and the tool contact data is used to estimate the associated robot attitude.The robot’s joint angles are derived from the corresponding posture matrix by solving the inverse kinematics,and the robot’s grinding route is created by connecting the tool contacts in series.Finally,the various processing stages of the system are integrated and developed into a single working platform to improve the reliability and automation of the system.