Experiment Based Structural Stiffness Calibration of a Virtual Robot Model

Abstract

The main advantage of industrial robots is their flexibility, if compared to CNC machines, and speed and accuracy, if compared to humans. To improve accuracy of industrial robots, their stiffness, as one of their most important structural properties needs to be determined. As stiffness of robots with rotary joints is pose dependent, it is convenient to perform FEM simulations to predict a robot's structural stiffness in a desired pose. However, simulation results may deviate in magnitude and trend if compared to measurements. The reasons for deviations are inexact model geometry, friction, component wear and other unknown effects. Our study focuses on structural stiffness measurements for FEM model calibration. A FEM model, with all unknown effects integrated in rotational stiffness' of the robot's joints is created and for each joint, rotational stiffness as a function of joint angles is determined. The calibrated model is used to generate a surrogate model for the robot's structural stiffness.