Learning EMG control of a robotic hand: towards active prostheses

Abstract

We introduce a method based on support vector machines which can detect opening and closing actions of the human thumb, index finger, and other fingers recorded via surface EMG only. The method is shown to be robust across sessions and can be used independently of the position of the arm. With these stability criteria, the method is ideally suited for the control of active prosthesis with a high number of active degrees of freedom. The method is successfully demonstrated on a robotic four-finger hand, and can be used to grasp objects. I. INTRODUCTION The focus on the properties of prosthetic robotic hands is currently undergoing a slow but steady shift. Whereas the cosmetic aspect traditionally was and still remains an impor- tant issue for most patients, regained dexterity becomes more prominent as advances in mechatronic systems offer such pos- sibilities. Consequently, most prosthesis manufacturers (e.g., Otto Bock, Motion Control, Liberating Technologies), besides their line of cosmetic hands, also offer a line of active hands. Such hands are controlled over myoelectric (EMG) interfaces, measuring the activity of the patient's muscles (viz. their motor units) in the lower arm. Such active hands however suffer from low acceptance among patients with amputated hands. One reason for this is the limited dexterity that such hands offer to the patient: with only one active degree of freedom, the hand can only open and close on command, and can thus only be used for very coarse grasping and handling. Whereas already many highly integrated prosthetic hands are under development or at production level (e.g., the DLR- HIT hand, the Cyberhand, the Fluid hand), control of such hands using noninvasive interfaces is still problematic and cannot be used on patients. Limitations lie with the number of fingers that can be controlled, session independence without retraining, and insensitivity to other arm muscle use, e.g., w.r.t. arm pronation or supination. In this article we present a highly integrated approach to the use of EMG recording of the human lower arm in order to control the opening and closing of three fingers of the hand. Although user independence is problematic, we do attain optimal results with respect to session independence, allowing a patient to wear use the prosthetic device in the morning without having to retrain it. Also we demonstrate the ability of using the device independent of the position of the arm. The method is demonstrated on a robotic four-finger hand, and can be used for grasping objects.