Myoelectric signal source separation for prosthetic control

点击翻译按钮获取中文摘要

This research focuses on the functional replacement of the wrist and hand, in the context of an individual who has an amputation (or congenital deficiency) below the elbow.  This represents almost two-thirds of the amputee population.   For those with upper limb deficiencies, currently available prosthetic limbs offer limited function.  This is due, in large part, to the lack of control information that can be delivered to the device by the user.  Significant advances have been made in recent years in interpreting a user's intent by monitoring the myoelectric signal at multiple sites, and using advanced signal processing to determine the intended motion.  The applicants have demonstrated that it is possible to control wrist and hand function with high reliability using a pattern recognition approach.   Although this represents a significant advance, pattern recognition allows only one function to be actuated at a given time.  This requires a user to manipulate one joint at a time; if a task requires multiple joints, these must be done in sequence.  Clearly this is not how one uses an intact limb.  This research promises simultaneous and independent control of multiple joints to deliver natural, coordinated motion to prosthetic devices.  This will be accomplished using a combination of source separation methods and inverse modeling to resolve the activity the forearm muscles.  By determining the level of activity of individual muscles which act as the "prime movers" in hand and wrist manipulation, each joint may be controlled independently. The ability to control hand and wrist function in a natural, coordinated manner will represent a tremendous enhancement in prosthetic control.