Brain-muscle-computer interface using a single surface electromyographic signal: Initial results

We review progress towards developing a new human-computer interface that uses only a single sEMG sensor to achieve complex control of external devices. The interface relies on an underlying neuromuscular result, in which we showed that the human neuromuscular system can simultaneously manipulate partial power in two separate frequency bands of a sEMG power spectrum at a single muscle site. Subjects are trained using visual feedback based operant conditioning. The two frequency bands can then be used as two separate control channels to achieve multidimensional control of external objects, using only a single measured sEMG signal. Our first results showed that subjects could hit targets on a computer screen in two-dimensions with a cursor. Cursor manipulation was then paired with a finite-state machine to achieve control of simulated and actual power wheelchairs. Work on brain-muscle-computer interfaces is still in its infancy, and fundamental questions must still be answered in terms of training, usability, and underlying neurophysiological mechanisms. However, our early results show promise that these interfaces may provide a new option to benefit the lives of severely paralyzed persons.