Improving Hole-Searching Accuracy for Peg-in-Hole Assembly With Manipulator Harmonic Drive Uncertainty

Abstract

Uncertainties of a robot manipulator include the inaccurate transformation between each coordinate, (e.g., robot base, flange, camera, tool, and workpiece) and the inherent variations within each links and mechanical components. In this work we consider the impact of harmonic drive uncertainty in peg-in-hole assembly. We investigate the accumulations of transmission errors within robot dynamic and nonlinear process in rotation. Models of manipulators with uncertainty parameters in harmonic drive systems are developed. The parameters in our harmonic drive model with uncertainty are identified through testing. The transformation metrics between each coordinate are also determined by detecting fiducial pattern and marker with well-calibrated camera. As a result we have a virtual robot model with parameters and uncertainties much closer to the real system. The simulation results show that the accuracy of hole-searching can reach 0.50 mm position error and 0.16 degree orientation error for hole locations that are not known as a priori.