sEMG-based estimation of human stiffness: Towards impedance-controlled rehabilitation

Abstract

In rehabilitation robotics, surface electromyography (sEMG) is extensively used as a human-machine interface, mainly for prosthetic/orthotic control purposes. The technique has been proved to be a highly accurate way of detecting a human subject's intended position, force and torque configurations. Widely applied in the clinics, it is gaining even more momentum as polyarticulated, ever-more dexterous self-powered rehabilitation artifacts appear on the market. In this paper we present a preliminary result about the usage of the same technique to estimate a human subject's hand stiffness in the presence of force-feedback. A novel force feedback control concept based on the modulation of the robot arm stiffness according to the estimated hand stiffness is presented. Thus, the robot arm is set to mimic the stiffness properties of the subject that is controlling the arm, in real time. Six intact subjects were immersed in a simple teleoperation task, in which force feedback was present; the hand stiffness was measured via force perturbation at the master's manipulandum and associated with the sEMG signals. This live estimation of stiffness was then used to control the impedance of the slave. Experimental results show that this system leads to high positional precision but high contact forces when the estimated stiffness is high, and vice-versa. The system has potential applications in impedance control of rehabilitation devices such as, e.g., upper / lower limb prostheses, self-powered orthoses and exoskeleta, leading to an ever-better integration with patients.