Kinematic Calibration of Parallel Robots Based on Total Least Squares Algorithm

Abstract

A new method for calibrating parallel robots is proposed. It requires measurements of all legs and the pose information provided by coordinate measuring machine. A total least squares minimization is employed to determine the kinematic parameters. The parameters of the measurement errors in the leg sensors are considered during kinematic calibration modeling and parameter identification program. Experimental results presented demonstrate that the root mean square pose error can be improved at 80% with the 42 identified parameters.