A Numerical Algorith to Identify Independent Grouped Parameters of Robot Manipulator for Control

Abstract

Many of modern control schemes for a robot manipulator require accurate information on the robot dynamics. Thus, it is critical to find the precise dynamic parameters, which include inertial parameters as well as dimensional parameters, accurately. One critical issue in determining accurate model of a robot manipulator is to estimate the inertial and dimensional parameters precisely, such as the mass, location of the center of gravity, and moment of inertia for each link. In order to use linear regression model, which is more efficient than a nonlinear optimization scheme, it is necessary to find a set of independent grouped parameters. Many of them are a product of an inertial parameter and a dimensional parameter or its squared value. Trying to find them analytically or by hand is a time-consuming and error-prone process. This paper proposes a numerical method to identify a set of independent parameters efficiently based on the dependency analysis of the regressors based on measured motion trajectory data of a robot manipulator and shows how they could be used in computing the required joint torque for a desired trajectory. The proposed scheme was applied to a 6-DOF robot manipulator, each link of which has ten inertial and dimensional parameters. The simulation results showed that the proposed method works well in identifying the independent grouped parameters and the estimated torque based on them were accurate.