A svd-least-square algorithm for manipulator kinematic calibration based on the product of exponentials formula

Abstract

In recent years, a great deal of research is conducted to improve the accuracy of manipulator kinematic calibration of which the product of exponential formula (PoE) is used to represent the manipulator kinematics, whose purpose is to solve the singularity problem of the Denavit-Hartenberg (D-H) parameters. However, the noise sensibility is still an open problem since a matrix inverse calculation of Jacobian matrix is inevitable during the process of solving the kinematic-linearized-equations to obtain the calibrated parameters. This problem may causes non-convergence, or low-accurate solution of calibration algorithm if the environmental noises and the error of endeffector’s actual frame measurement techniques are considerable. This paper presents a kinematic calibration method using singular value decomposition least square algorithm based on the product of exponentials formula (SVD-PoE-least-squares algorithm) to improve the accuracy of calibrated parameters. The proposed algorithm is evaluated in simulation level using a 6-DOF puma 560-type manipulator. The obtained results have shown that SVD-PoE-least-square algorithm is insignificantly affected by environmental noises, and, the proposed method can complete the robot calibration with respect to the work frame directly.