Traditionally, the repeatability of the industrial robot is an important index as evaluation ofthe robot accuracy. But with the application of the robot more and more widespread, manyoccasions require a higher positioning accuracy of the robot, such as robot vision and offlineprogramming. Thus, the positioning accuracy of the robot is attracted with more and moreattention by many researchers.This thesis focus on the kinematic calibration of a2-DOF translational parallel manipulatorcalled X2in order to improve the absolute positioning accuracy of the robot finally, mainlyincluding the following aspects:The ideal kinematic model is established and the workspace of the robot is calculatedaccording to some known conditions. The input and output error model with actual geometryerrors of the robot is then established through the error analysis. The sensitivity of eachgeometric parameter error on the end position error of the robot is also calculated throughthe actual model derivation.The linear error model is established using the differential method and further the parameteridentification model is established by the least squares method. The computer simulation ofthe robot calibration is accomplished using MATLAB tools to prove the correctness andvalidity of above model and method.Compared with the differential method, another linear error model is established directlyusing vector perturbation method and further the parameter identification model is alsoestablished. Similarly, the computer simulation of the robot calibration is accomplishedusing MATLAB tools.The robot calibration motion software is designed based on Labview developmentenvironment and NI-motion library to complete the moving of25measuring pointsautomatically, and also including a single-axis adjustment, back to zero and other auxiliaryfunction modules.The measurement of the robot end position error is completed by the absolute tracker andreflective balls with motion software. The two groups of the parameter identification basedon different models and methods were solved. Then, the robot end position errors aremeasured again after error compensation and the results of the two methods were comparedeach other. |