Abstract:
The prosthetic arm market is small; perhaps 70,000 people lack arms in
the US. This, coupled with the difficulty of making and controlling a
mechatronic hand replacement, has forestalled innovation in arm prostheses
in the fifty years since its introduction. Since the Boston Arm project in
1965, the popular press has promised thought-controlled prosthetic arms,
yet the promises of scientific research have not often been kept in the clinic.
[1] We aim to surmount these obstacles by developing an open myoelectric
signal processor targeted at researchers, hobbyists, and video game
enthusiasts. Our device is capable of processing 16 channels of surface
electromyographic (EMG) (or other data), applying pattern recognition
algorithms in real-time via a power-efficient Blackfin Digital Signal Processor
(DSP), and delivering the results through ethernet, I2C, RS232, USB, and
Lego NXT bus. We hope that this open, commodity level platform will
become a disruptive technology, encouraging experimentation in the
algorithms and applications of a field that has been sequestered too long in
the research lab.
