Research On The Method Of Vision-based Calibration Of AUBO Robot

Industrial robots are increasingly used in light-load,high-speed,high-precision industries,and especially in the high-precision industries such as assembly,dispensing,cutting,and aerospace manufacturing of 6-DOF industrial robots.The higher requirements are put forward,and the development of an industrial robot calibration method with independent intellectual property rights is of great significance.In this paper,the six-degree-of-freedom AUBO industrial robot is taken as the experimental object,and the end positioning accuracy is improved.The geometric parameters and comprehensive error compensation method are used to calibrate the end positioning accuracy,and the calibration test is performed.Aiming at the shortcomings of the existing measurement methods,a vision-based measurement method was proposed.The kinematics model of the AUBO robot was established.The mathematical relationship between the end pose and the adjacent coordinate system was deduced.The inverse solution of the AUBO robot was obtained by the mathematical analysis method.The simulation model was established using MATLAB robotic toolbox.The simulation results and the position and attitude of the teach pendant were exactly the same,which verified the correctness of the established model and the correctness of the inverse solution.Aiming at the problem of insufficient accuracy of Zhang's camera calibration method,an improved sub-pixel Zhang calibration method was proposed,and a standard checkerboard was used for comparison test.The experimental results showed that the camera calibration accuracy was improved by 0.12 pixels,verifying the high precision of the proposed method.The eye-in-hand fixation method of Eye-in-hand was used to establish the basic equations for hand-eye calibration of the robot.The hand-eye calibration matrix was obtained using the matrix straight-product method.In order to identify the geometric parameters of the robot,the robot's position measurement experiment was performed.The kinematics parameters of the robot were identified by multiplication.The feasibility of the visual calibration method and the advantages of higher accuracy were verified by the laser tracker based on the distance error model.In view of the complexity of the end positioning error and the limitations of kinematic calibration,the BP method is used to predict and compensate the terminal integrated errors.Based on this,the PSO-BP network's end pose prediction and compensation algorithm is adopted.In MATLAB,the simulation model of the end position and posture prediction and compensation is established.The simulation results show that the average distance error at the end position is increased from 0.635 mm before calibration to 0.294 mm after calibration,and the accuracy is improved by 53.7%.This proves the calibration method.Correctness and feasibility.This study provides a reference method for the calibration of robot geometry parameters and integrated errors.