Jixie chuandong (Jan 2019)
Research on Joint Stiffness Identification and Error Compensation of the Serial Six DOF Robot
Abstract
To improve the absolute positional accuracy of the serial six-DOF robot， the joint stiffness error of industrial robots after geometric parameter error compensation is studied. Firstly，the one-dimensional joint stiffness error model of industrial robots is established based on the virtual joint model. Secondly，in order to improve the identification accuracy and efficiency of joint stiffness parameters, the BP neural network is applied to fit the stiffness error model to optimize the initial population fitness of genetic algorithm. Finally，the laser tracker AT930 and ER10L-C10 robot are used to verify the above error model and joint stiffness parameter identification algorithm. The experimental results show that the average distance error and maximum distance error of the robot are 0.248 5 mm and 0.333 2 mm respectively after the joint stiffness error compensation. Compared with the distance error before error compensation，the positional accuracy of robot is improved by 33.7%. Therefore，through the proposed improved genetic algorithm can identify the joint stiffness parameters accurately and improve the positional accuracy effectively.
