Stable admittance control without inverse kinematics

Abstract

Admittance control is mostly used for human-robot interaction. When the admittance control is applied in the task space, it needs the Jacobian matrix. When the admittance control is in joint space, it requires inverse kinematics of the robot. In this paper, the admittance control is modified to only use the orientations of the end effector to generate the desired joint angles to avoid inverse kinematics. The classical PD control is also modified with adaptive compensation and sliding mode compensation. The stability of these two PD controllers are proven. Real time experiments are presented for with a 2-DOF pan and tilt robot and a 4-DOF exoskeleton. The experiment results show the effectiveness of the proposed controllers.