Research on the kinematic parameter calibration method of industrial robot based on LM and PF algorithm
Xin Deng, Lianzheng Ge, Ruifeng Li, Zhiheng Liu
Abstract
The absolute positioning accuracy of the industrial robot is one of its important performance indexes, which is impacted by the key factor of robotic kinematic parameters. Therefore, based on the MDH model a calibration method of robot kinematic parameters, which combines the Levenberg-Marquardt algorithm with the particle filter algorithm is proposed. Firstly, the MDH model of an industrial robot is established, and the parameters in the tool coordinate transformation are also regarded as the parameters to be calibrated. Then the end error model is established. Secondly, the initial optimization is carried out using the Levenberg-Marquardt (LM) algorithm. Meanwhile, the particle filter algorithm is used to further optimize the parameters considering the measurement error. Finally, compared with other methods, such as spatial circle fitting method, least square method and extended Kalman filter algorithm. The results show that the kinematic parameters of the robot are accurately calibrated and the absolute positioning accuracy of the industrial robot is significantly improved by this method. Compared with other methods, the parameters calibrated by this method have stronger generalization ability.