Return to the archive index

Re: Character Input w/o using Hands, Voice etc.

"R. Paul McCarty" <> writes:
> The other possibility is to measure the electric signals from the
> muscles of the hand, or some other part of the body and you could just
> twitch your muscles almost imperceptably and convert it into a chorded
> input.

You a leaning toward the dead subfield of electromyography, "Single
Motor Unit Training".  A motor unit (SMU) is a motorneuron and the
associated muscle fibers.  Using simple biofeedback, humans can learn
to control individual motorneurons.

The bandwidth potential is very good.  For each SMU you can reliably
vary the firing rate over a range of 5-20 pulses per second (pps), to
within 1pps.  That's about 4 bits.  You can reasonably expect to
change your firing rate a few times per second, or about 12 baud per
SMU.  I can isolate about a dozen SMU's from a single set of surface
electrodes for an aggregate bandwidth of 144 baud.  Assuming a 6 bit
encoding, we get 24cps, or about 240 wpm--comparable to the very
fastest typing.  Variable length encodings improve the situation even
further.  There are about 750 SMU's close enough to the surface to be
easily trained.  That's a total bandwidth of 9kbaud...

This work was originally done in the late 60's and early 70's by John
Basmajian and others.  The field seems to have died out in about 1975.

I've gone as far as reproducing the analog experiments from the 60's.
The apparatus is quite simple, and fairly cheap--but the electronics
are fiddley.  I have ideas for the digital processing, but have not
spent the months of effort to develop the necessary algorithms.

Drop me a note if you'd like bibliographic information.
-- 
La Monte H.P. Yarroll      Home:                
   If you remember nothing else:           NIC Handle: LY
   GPL - "Just give source a chance."

From Wear-Hard Mailing list Archive (WH)
Maintained by R. Paul McCarty

Archive created with babymail