Kinematic Calibration of Parallel Robots Based on Least Squares Algorithm

Dayong      Yu; Qiang      Zhang

doi:10.2174/2212797611104030226

Abstract

A calibration method of parallel robots for semi-physical simulation platform of space docking mechanism is presented in this paper. To calibrate the parallel robot, a coordinate measuring machine to measure the distance between three standard spheres on the moving platform and three standard spheres on the base platform is attached to the base platform and the pose calculation is established by the distance information. By treating each sub-chain of the parallel robot as a joint-link train, the pose error model in which kinematic parameter errors are considered is established from the kinematic equations and the loop properties of parallel mechanism. Using least squares and iterative algorithm the solution of kinematic parameters has been realized. The factors that influence the accuracy of kinematic calibration are studies in detail. The corresponding simulation calculation demonstrates that to find out a coefficient matrix with a good condition number is a precondition of the calibration algorithm. Some examples and calibration experiment are given to demonstrate the effectiveness of this approach. The position error RMS of the parallel robot is reduced from 4.2mm to 0.4mm and the orientation error RMS is also reduced from 0.6° to 0.1°. In this article, we have discussed some recent patent on kinematic calibration of parallel robots.

Keywords: error compensation, error modeling, kinematic calibration, least squares algorithm, parallel robot, pose accuracy, inverse kinematics, simulations, forward kinematics, kinematic model