A novel kinematic calibration algorithm for reconfigurable robotic systems

Abstract

A modular reconfigurable robot system is a collection of individual link and joint components that can be assembled into different robot geometries for specific tasks requirements. However, the machining tolerance and assembly errors at the module interconnections may affect the positioning accuracy of the end-effector. Based on the product-of-exponentials formula and recursive forward dyad kinematics, this paper describes a novel kinematic calibration algorithm for modular robots. The error correction parameters are assumed to be in the relative initial positions of the dyads. A six-parameter calibration method is derived on the ground of a linear superposition principle and differential transformation theory. An iterative least square algorithm is employed for the calibration solution. A simulation example of calibrating a three-module manipulator is demonstrated. The result has shown that the average positioning accuracy of the end-effector increases two orders of magnitudes after the calibration.