A calibration method of industrial robots based on ELM

A calibration method for enhancing absolute position errors of industrial robots is presented by the paper. In this study, the ELM algorithm is utilized for error prediction. Firstly, the relationship between absolute position errors and robot joint angles is built by training samples. Then joint angles of positions to be predicted are provided to predict position errors based on the relationship. Besides, this paper compensates the position coordinates of the predicted positions. All processes implemented in the paper needn't to build the kinematic model. Finally, the experiment is carried out on the KUKA KR210 R2700 industrial robot. Experimental results indicate that the maximum of the spatial position errors is from 1.6866 mm to 0.3565 mm with the reduction of 78.86%, which verifies the method's effectiveness.