Design of a RLS-Based Real-Time State Observer for Estimating Joint Accelerations and Inertias of a Robot Manipulator

Abstract

In this paper, a RLS-based real-time state observer for a robot manipulator is designed. The availability of both the inertial mass and acceleration information in real time provides control efficiency to design robust control methods such as time-delayed control. Since it is difficult to model the inertia and to measure the acceleration accurately, an RLS-based joint model observer (RLS-SOB) is designed to estimate the inertial mass of each joint as well as angular acceleration in real time as the configuration of 2nd order linear model. The proposed method is empirically verified for a robot manipulator.