Writing skills transfer from human to robot using stiffness extracted from sEMG

Abstract

Studies of human motor behaviors have shown that central neural system (CNS) is able to adapt force and impedance in order to optimally interact with interactive environment. Muscle activities regulated by CNS can be represented by surface electromyography (sEMG) measured by electrodes attached on the skin. Inspired by the idea that muscle impedance adaptation reflects motion skills, sEMG based human-robot skill transfer, in particular, the writing skill transfer has been developed based on impedance adaptation extracted from sEMG. Squaring and low-pass filtering based signal envelop extraction algorithm and as well as re-sampling method are employed to extract incremental smooth stiffness from sEMG signals which is then transferred to robot to mimic human motor behavior. The effect of the proposed sEMG based bio-control is evaluated by writing task in comparison with constant stiffness control. Results show that sEMG based human skill transfer has significant effectiveness for skills transfer between human and robot, and it has a great potential to be used in teleoperation.