Analysis And Compensation Of Vision Tracking Measurement Of The End Position And Pose Error In Robot Milling

Industrial robots have lower absolute trajectory precision in motion due to mechanical structures and errors in manufacturing of components and assembling of the whole set.In addition,the end precision of industrial robots is highly pose-dependent,and varies considerably with different poses in the reachable space.When the robot is applied to the milling task of large complex parts,the milling precision will be unevenly distributed,and the trajectory position error in degree of millimeter seriously affects the milling quality of the workpiece.Research on the measurement methods of the end position and pose error in the robot's milling trajectory,analysis of the error distribution law,and proposal of an error compensation strategy are essential to improve the precision of the robot's milling processing.Aiming at the absolute position and pose error of the robot in milling,a method for calculating the end position and pose error with binocular vision system in tracking and measurement is proposed.A binocular vision tracking measurement platform for robotic milling processing was built,and coded target points were configured.The coding target points were dynamically identified through the binocular vision system to complete the construction of the coding target point coordinate system.Based on the pose information of the robot,two calibration algorithms for the spindle coding target point coordinate system are proposed,and the mapping relationship between the spindle coding target point coordinate system and the tool coordinate system of robot end is established.Through the conversion between the coordinate systems,the calculation is realized of the actual pose error and position error of the tool end during the robot movement.The application of coordinate system calibration algorithm,pose error measurement algorithm,and position error measurement algorithm in robot milling processing is realized.Based on the algorithm principle,pc software is developed.The communication between the software,the robot and the binocular vision tracking system is completed in the coordinate system calibration module,and the accuracy of the two coding target point coordinate system calibration algorithms is compared through test experiments.The feasibility test of the pose error measurement algorithm and the position error measurement algorithm is carried out through test experiments,and the accuracy of the error measurement is verified by a laser tracker.Finally,aided by the development software,the trajectory position error is measured in the robot milling experiment,the results of which prove the validity of the algorithm.The BP neural network model optimized by the PSO algorithm is trained to simulate the inherent relationship between the robot position and pose and the end position error in its milling trajectory.A position error prediction model based on neural network is established,a measurement area is selected within the reachable space of the robot,and the position error collection plan of the TCP with different trajectories is formulated to collect sample data.The 14-fold cross-validation method was used to test the established network prediction model,and the optimal structural parameters of the model were determined.The network prediction model was optimized through the PSO algorithm,and the final prediction model of the robot end position error was established,and its prediction accuracy was verified.The characteristics of the end position error distribution within the robot measurement space are analyzed,and the Z-direction position error distribution is displayed for the common milling plane within the measurement space.An error iterative model based on the chord-cut method is established and introduced into the robot's axial position error compensation process.A comprehensive error compensation strategy is proposed that combines the robot end position error prediction model and the error iterative algorithm.This comprehensive compensation method is used to calculate the compensation amount of the robot milling trajectory,and an offline compensation experiment is carried out.The position error compensation is realized by correcting the TCP absolute trajectory,and the maximum depth of cut error of the workpiece is reduced by 80%.Based on the above research,the comprehensive error compensation method is applied to the robotic milling processing of large marine propellers and complex spacecraft cabins,which significantly improves the machining precision.