The Research And Simulation Of Calibration Method In 6DOF Serial Robot

With the development of modern economy, industrial robot, computer systems and programmable controller play an important role in factory automation, the pose accuracy of the end of robots is more and more demanded, absolute positioning precision of the robot is particularly important to the exchangeability and off-line programming technology. Robot calibration is one of the key technologies in the practical off-line programming, the calibration is to identify the exact parameters of the robot model with the application of advanced measurement methods and parameter identification method based on model, so as to improve the process of robot absolute accuracy. The main works of the dissertation contain:(1) Robot kinematic model and deviation analysis. Taking the IRB140 industrial robot as reference, using the D-H modeling method, establishing the robot rod-coordinate-system, analyzing the robot kinematics positive and inverse solutions, while solving the robot Jacobian matrix so as to provide the theoretical basis for the following analysis of the deviation model; Analyzing deviation sources that affects the accuracy of robot, distinguishing the repeat accuracy and absolute accuracy.(2) The deviation model and parameter identification deviation of robot. Based on the kinematic equation, the calibration model of deviation was deduced which is suitable for calibration, the calculating program of error model was written with MATLAB language, and the reliability of the error model is verified through experiment simulation; The kinematic parameter errors were identified and simulated by applying the least square method and maximum likelihood estimation method respectively, and the simulation results of the two kinds of parameter identification were compared; finally the general conclusion of the parameter identification performance was given.(3)The research and simulation of robot inverse-calibration based on neural network. In order to overcome the defects of calibration only for geometrical parameters of the robot, the inverse kinematics calibration method based on neural network was put forward, that attributes all the errors to the joint angle error, the corresponding joint error value of each angel was calculated through a certain algorithm, then the joint angel and the corresponding joint error are used as input and output to train the neural network, the calibration accuracy was improved through the joint angle compensation, and the simulation results prove that the neural network optimized by genetic algorithm can further improve the calibration accuracy.(4) Research of robot-controlling with gravity compensation. The dynamic model was established with Lagrange method, since the gravity casts significant influence to the dynamic features of the robot, by casting gravity compensation method, the robot controller can overcome gravity and its stability can be analyzed, so the joint error was reduced, the robot can reach it’s desirable position; The simulation through MATLAB shows that the robot with gravity compensation can reduce the error of joint and realize stable control.