Research On Kinematic Calibration And Error Compensation Of Industrial Robot

With the ongoing upgrading of automation production lines across the world, as oneof the main means of achieving methods for automation production lines, industrial robotshave been widely used in industrial production, such as automobile body painting, laserwelding, palletizing and assembly. Absolute positioning accuracy of industrial robots ismillimeter, to enhance the versatility of the industrial robots, it is extremely significant andimportant to do some research on calibration technology of industrial robot to improve theabsolute position accuracy.Based on the actual structure of MOTOMAN UP20robot, given the shortcomingsand deficiencies of DH model, modified DH model was adopted to establish thekinematical model of UP20robot, and the kinematics solution and inverse kinematicssolution was given. On the basis of the comprehensively analysis of industrial robot error,error model based on distance accuracy was established to lay the theoretical foundationfor subsequent parameter identification and error compensation.On analysis of the pros and cons of the current methods used to solve the inversekinematics, a new method using the results of BP network as a criterion was proposed forsolving the inverse kinematics solutions. Based on the robot forward kinematics solution,the BP network was trained by using the results of forward kinematics as sample data,predicting result was used to compare with the solutions got by algebraic method andmean square error was used to choose the very candidate of algebraic results. Forwardkinematics and inverse kinematics solution calculation program based on MATLAB GUIwas worked out to enhance the convenience.To avoid the problems on direct measurement of industrial robot end-effector postionand posture, error model based on distance accuracy was adopted for parameteridentification. Structure lightening3D measuring technology based on digital raster wasused for measurement positon and posture data collection of end-effector in work space. By using Geomaic Studio for point cloud data processing, the structural parameter errorsof kinematical model was obtained by least squares method and commad coordinateposition after joint error compensation was calculated by Newton-Raphson method.Finally, backward error compensation was realized by using moving window methodfor modification of slicing trajectory of aircraft skin mold. Revised slicing trajectory wasused for rapid manufacturing of aircraft skin mold.