A vision-based fully-automatic calibration method for hand-eye serial robot

Abstract

Purpose – The purpose of this paper is to present a fully automatic calibration method for hand-eye serial robot system is presented in this paper. The so-called “fully automatic” is meant to calibrate the robot body, the hand-eye relation, and the used measuring binocular system at the same time. Design/methodology/approach – The calibration is done by controlling the joints to rotate several times one by one in the reverse order (i.e. from the last one to the first one), and simultaneously take pictures of the checkerboard patterns by the stereo camera system attached on the end-effector, then the whole robot system can be calibrated automatically from these captured images. In addition, a nonlinear optimization step is used to further refine the calibration results. Findings – The proposed method is essentially based on an improved screw axis identification method, and it needs only a mirror and some paper checkerboard patterns without resorting to any additional costly measuring instrument. Originality/value – Simulations and real experiments on MOTOMAN-UP6 robot system demonstrate the feasibility and effectiveness of the proposed method.