Abstract: The geometric parameter error of robot leads to its poor absolute positioning accuracy, which restricts its application in off-line programming. The joint stiffness of the collaborative robot is small, so its flexible deformation is largely affected by its self-weight and end load, which further reduce its positioning accuracy. Aiming at this problem, the kinematics calibration method of adjacent three-axis parallel(2-4 axis) cooperative robot is studied, and the corresponding calibration software is developed to apply the method to engineering practice. Firstly, a nonlinear geometric parameter identification model is proposed, and the redundancy of the model parameters is analyzed by using the inverse function principle of the model's nonlinear equations. Further, a joint stiffness identification model is established. Considering the mutual interference between geometric parameter identification and joint stiffness identification, a comprehensive identification method based on iterative thought is proposed. The calibration experiment of UR5 robot is carried out using this method. The experimental results show that the absolute positioning accuracy of the robot under no-load and load after calibration is improved by 75.5% and 85.1%, respectively, compared with that before calibration, which verifies the effectiveness of the calibration method in this paper.

Key words: collaborative robot, kinematics calibration, nonlinearity, geometric parameter, joint stiffness