Dynamic Parameter Identification based on the Least Squares method for a 6-DOF Manipulator

Abstract

Accurate dynamics play crucial roles in designing advanced control algorithms to ensure the feasibility, stability, and efficiency of the system. However, the manipulator is a complex multi-input-multi-output system, and so many system noises seriously affect parameter identification results, thereby the process of determining them is challenging. In order to manage this challenge, a Least Squares (LS) method is proposed to estimate the dynamic parameters. First of all, the kinematics of the system is built according to Denavit-Hartenberg (DH) notation, and the dynamic model is calculated by using Lagrange-Euler equations. After that, the regrouped parameters in the dynamic model are given to the general linear matrix to apply the least squares method for the model. Finally, to demonstrate the effectiveness and reliability of the proposed method, data acquisition is carried out by using a PD controller in simulations via Simscape Multibody/MATLAB Simulink for the dynamic estimate model and the Solidworks robot model. After that, the Root Mean Squared Error (RMSE) formula is used to analyze the accuracy of the LS method.