Kinematic Parameter Calibration Method of Light Weight Robots Based on Sequence Quadratic Programming Algorithm

Abstract

Kinematic parameter calibration can improve the absolute positioning accuracy of the robot tool center point (TCP). However, the current least-squares (LS) iterative method for calibration is sensitive to noise and ineffective when the Jacobian matrix is close to singular. In addition, the robot kinematic parameter errors cannot be directly compensated because its structure needs to satisfy Pieper's criterion. Thus, a kinematic parameter calibration based on sequence quadratic programming (SQP) algorithm is proposed to solve the above two issues. Firstly, the kinematics and its error model of TCP positions are established. Then, the kinematic parameter errors are identified by SQP algorithm. Finally, the identified kinematic parameter errors are converted into joint angle errors for error compensation. The simulations and experiments are carried out on the self-developed light weight robot. Simulation results show that the proposed calibration method is insensitive to noise. Experimental results reveal that mean positioning accuracy has increased 88.17% and 63.58% than that of the before calibration in sampling and verification regions, respectively.