Real-Time Myoelectric Control of a Virtual Upper Limb Prosthesis via Lower Leg Gestures: Preliminary Results

Abstract: Few options are currently available for high-level upper limb amputees to control a powered prosthetic arm. Existing techniques are either invasive, requiring surgery and recovery time (approximately one year), or are complex, requiring significant training time (several weeks). We have developed a noninvasive technique where lower leg and foot gestures recognized by surface electromyography (EMG) map to homologous movements of the prosthetic arm and hand. The mapping is based on the alignment of the degrees of freedom of the wrist and ankle in addition to the similarity of finger and toe movements (grasps), so each of the mapped gestures are intuitive and the entire mapping takes less than five minutes to learn. Our previous work has demonstrated that 10 or more lower leg and foot gestures can be recognized offline via surface EMG, and we now present the results of a real-time experiment in which subjects controlled a simulated prosthetic arm using this technique. The Target Achievement Control (TAC) test was used as an experimental task to evaluate subjects' ability to move the virtual arm to target postures offset from the neutral posture in several degrees of freedom. Common performance metrics for this task are presented and compared to the case in which subjects used arm gestures, recorded via an analogous surface EMG configuration on the arm, to perform the same task.