Research On Kinematics Calibration And Error Compensation Of Industrial Robots Based On Distance

Industrial robots are now growing rapidly as an important part of intelligent manufacturing.Traditional applications only require industrial robots with high repeatability.But there are quite high requirements for the absolute pose accuracy of industrial robots in flexible production lines,off-line programming,product quality control,and so on.Kinematics calibration is one of the most important way to improve the pose accuracy of a robot.Nowadays,it is often necessary to obtain complete robot end position information during the calibration process.This requires the use of expensive,complex equipment to measure the end of the robot accurately,which is costly and inconvenient.In this paper,a distance-based kinematics calibration method for industrial robots was proposed.A distance-based draw-wire calibration system was developed.The developed calibration method and system were verified by experiments.In this paper,a general 6 DOF industrial robot was taken as the research object,and its kinematics model based on MDH(Modified Denavit-Hartenberg)was established.The correctness of the model was verified by comparing the numerical simulation results of the kinematics model with the actual robot positions.Based on the analysis of the kinematics parameter errors of the robot,the position error model of the robot was established.Based on this,the distance error model and the distance-based error regression model were further established.The coupling relationship between kinematics parameter was analyzed.These work laid the theoretical foundation for kinematics parameter identification.This thesis explored the pre-identification method of the reference point of the measuring equipment in the distance-based parameters identification algorithm.The kinematics parameter identification was used as the nonlinear parameter optimization problem.According to the algorithm framework classification,a traditional linear optimization algorithm——Gaussian Newton method,an intelligent search algorithm-AFSA-PSO algorithm and a Trust-Region algorithm were used to study the parameter identification problem.The three algorithms were simulated and compared.The numerical simulation results showed that the Trust-Region identification algorithm and the Gauss-Newton method had the most accurate simulation results,and the AFSA-PSO algorithm had the second simulation effect,but it could also be accepted.A draw-wire robot calibration system was developed based on the modeling and parameter identification methods.They include the selection of the draw wire sensor,the structural design of the draw wire adapter and the guide mechanism,the software and hardware development of the data acquisition card,the development of data acquisition software and calibration softwares.The calculation of the effective draw wire length in the calibration system was discussed.Finally,the robot was calibrated using the draw-wire robot calibration system.Three kinds of algorithms were used to identify the parameters and compensate the parameters to the controller of the robot.Then the accuracy of the calibrated robot was verified by the laser tracker respectively.The actual calibration results of the three algorithms were analyzed.The experimental results indicated that the absolute pose accuracy of the robot was greatly improved by using the AFSA-PSO algorithm and the Trust-Region algorithm.In contrast,the Trust-Region algorithm was more convenient to use,more efficient,and had the best identification accuracy.The absolute pose error of the robot was finally reduced from 0.928 mm to 0.549 mm using the Trust-Region algorithm,and the pose accuracy was improved by 41%.The experimental results verified the reliability of the developed draw-wire robot calibration system and the effectiveness of the parameters identification algorithm.