Physical Parameter Estimation for Feedforward and Feedback Control of a Robot Arm with Elastic Joints

Abstract

This paper presents a practical modeling for fast and exact trajectory control of a robot arm with elastic joints. We previously proposed a torsion-angular velocity feedback (TVFB) scheme for suppressing residual vibration of the tip of the robot arm using motor-side measurements only. The scheme utilizes a simple nonlinear observer based on a physically parameterized dynamic model of the robot arm with elastic harmonic-drive (HD) gears. In this paper, we combine a rigid-joint-model based feedforward scheme with the TVFB for executing exact trajectory control, since the implementation of an elastic-joint-model based feedforward is complicated. The TVFB enables to suppress the arm-tip vibration due to feedforward errors between the elastic and rigid joint models. In order to build the combined controller, we propose an accurate physical-parameter-estimation method based on both the linear and nonlinear least squares using the motor-side measurements only. Feedback and feedforward control experiments are conducted to validate the effectiveness of the modeling, estimation and TVFB using a serial two-link robot arm with elastic joints, which simulates the first two joints of a SCARA robot.