The purpose of our study is to achieve the dynamic model-based control of a nonlinear robot arm, while taking joint-elasticity into consideration. We previously proposed a multi-input multi-output system identification method, called the decoupling identification method, for a planar two-link robot arm with elastic-joints due to the Harmonic-drive reduction gears. The robot arm is treated herein as a serial two-link two-inertia system with nonlinearity. Physical parameters such as motor inertias, link inertias, joint-friction coefficients, and joint-spring coefficients of the robot dynamic model are accurately estimated by the decoupling identification method. This paper describes a simple method for achieving the vibration suppression control of a serial two-link two-inertia system based on physical-parameter estimation. We propose a torsional angular velocity feedback scheme, which can be “plugged-in” to an existing PI velocity controller, using a nonlinear state observer based on the accurately estimated dynamic model. In addition, we propose a gain-scheduling control scheme that involves switching inertia parameters to compensate for payload variation. Through several experiments, we demonstrate the effectiveness of the proposed control method by using the elastic-joint robot arm.
抄録全体を表示