Adaptive robust detection of below lesion, noninvasive electromyographic signals for muscle control

Currently, use of natural signals for control of a prosthetic/orthotic device is minimal, at best. Above lesion (above the spinal cord injury) control for such devices has been done successfully, but is bulky and slow. Such control is “unnatural” because the controlling muscles are being used for their own function at the same time. Below lesion control allows the controlling muscle signal to be isolated for a single task and is therefore “natural”, although below lesion electromyographic (EMG) signals have not been captured and used for control of prosthetic/orthotic devices. The EMG can be detected noninvasively (without breaking the skin), but the signal to noise ratio is low and the signal is extremely noisy. This paper demonstrates a nonlinear filter sequence which detects these EMG signals and which is robustly resistant to the occasional spikes present in the damaged neurological system. Moreover, the sequence adapts to the changing needs of the individual by allowing for muscle fatigue. Presently, robust controllers are being developed that will use this filtered signal to control a prosthetic/orthotic device