Hmm sounds like there are some nifty projects that are
afoot!
Here's my $0.02 on the subject.
Material dermal biocompatibility --
it's a semi-non-issue for most people for most common
materials, especially when the contact duration and
frequency is relatively minor (a few hours once in a blue
moon or several minutes intermittently).
The problem is just to avoid things that are highly
toxic, as well as things that are allergens to some minority
of the people. The situation is basically like jewelry,
but to a much lesser extent since you might wear a
watch, ring, necklace almost all the time, and things like
ear-rings are in such intimate contact (potentially through
a pierced hole in a very sensitive area) that they're almost
in the "implant" category.
Because the skin and tissue resistance is so high, and
because you'd be using high impedance (multi-megohm)
sensing (most likely), the contact resistance differential
is basically insignificant whether you were to use
lead, silver, aluminium, gold, platinum, or anything else.
The thing that would degrade contact reliability would be
the formation of thick layers of oxide / corrosion or grime,
or provoking a callous / inflammation that'd block the
electrode from having good clean skin contact.
So if you think about the materials and scenarios
commonly experienced with jewelry, various forms
of stainless steel, silver, gold, various plated forms of
those, et. al. you'll find both information
on corrosion / allergen / durability issues, as well
as possibly finding much cheaper and more common kinds
of material *sources* than things that are intended to
be medical electrodes! I'd probably just get some
good purity stainless steel 'washers' and start with that
as an experiment. It's (relatively) cheap / easy enough
to get various kinds of 'gold' (plate / alloy / whatever)
jewelry wires / bars / whatever if you decide you want
to work with those. Careful of the soldering, though,
obviously leaving semi-toxic metal or flux residues even
on a very inert platform wouldn't be good for the wearer!
You might try to do some electric spot / arc welds if
working with steel or platinum or something not so easily
soldered. For silver / gold alloys just use the kind of
either silver-solder or brazing material or MAPP / acetylene
type of welds and fillers that a jeweler would use on a
ring or necklace or something that would have skin
contact in the treated area (unless that'd be on the
reverse face).
Silverware, or electronic component leads may be other
sources of some pretty bio-inert electrode type supplies.
As noted previously, the saline gel can help with skin
resistance and intermittent contacts.
Electrode placement / geometry -- Well you could
do a bit of research on what people have done wrt.
biofeedback, TENS devices, polygraphy, et. al., and
consulting some basic anatomy books to see where the
involved muscles and their nerves run would be
informative.
However the old experimenters maxim applies well --
hook her up and see what you get! I expect if you
used something like an instrumentation amp feeding into
something like a PC based sound card or custom ADC
reading out into the game port / serial port / USB or
whatever you could easily do oscillographic plots of
the signal responses, geometric differential responses,
correlations or anti-correlations of responses between
differential point pairs (A-B) vs (C-D), etc. quite easily
when tinkering around with electrodes and their placement
long before you have to finalize the design of the
production system. I imagine that in a couple of hours
of tinkering you could easily identify the signal
shapes, amplitudes, and physiological actions to
selectively trigger several distinctly recognizable
sensor events. It shouldn't be NEARLY as complicated
as some of the more advanced EEG or even EKG type
monitors that are done routinely. And since you're
a part of the control loop you can always invent patterns
of movements that would be easily recognized by the
system as being artificial and control oriented
vs accidental twitch responses -- e.g. do the same action
left then right then left again in rapid succession or
'double-grip' (is this the new 'double-click?') quickly
or whatever works for your system.
Signal processing -- yeah I think just some garden
variety low voltage battery powered instrumentation
amps followed by some variable gain stage of x5 to x2000
should suffice for sensing pretty much anything the body
can easily emit wrt. neuro-muscular galvanic activity close
to the surface. Just feed it into an o-scope or PC
digitizer while you're playing around then work on whatever
more compact digitization and signal processing system
you may eventually want. I'd probably just stick an
mini-ITX or smaller type of low power PC into the thing
if I was building something of the size of an immersive
costume since space and a few watts of power aren't
too hard to come by and it's easier than doing 8051
embedded code for a quick hack.
I'd think it'd be easier to hook up about (at least)
eight sensors per forearm, giving several resultant
combinatorial differential pairs to look for cross correlations
against.. that should give you plenty of data to generate
several different signalling modes reliably.
If you want to get more complicated with the sensing
wrt. muscular activity you could always actively
look at things like sub-surface impedance to a low
power induced RF field like the body fat monitors do --
generate a signal that travels through the body,
different frequencies for different depths of penetration,
and look at the AC impedance vs frequency to see what
the tissue type / density is... And clearly when you
contract a muscle that'll change the position and density
of the tissue above / around it in a detectable way,
though really I expect that's not needed.
Other ideas -- I understand that a common technique in
TV type special effects is to have manual actuation of
one part of a given 'creature' and then programmatically
control the other motions of the creature relative to
believable patterns that follow suit the action of the
manually controlled parts, so you get composite behaviors
like 'looking around', 'smelling', 'hiding', 'dancing' or
whatever based on pre-programmed motion
control / behavior algorithms that extrapolate from the
manual inputs that are given to produce more intricate
and 'naturally flowing' results.
You might look at using things like face expressions /
muscles too if the hands / toes are too busy for much
control, those are easily picked off with electrical
galvanic sensors or strain gagues etc.
Good luck!
. I'm in my last year of electrical engineering technology, and we need to design and construct a final project. On the list of possible topics was the idea of a bio-med game interface. It would be biomed probes detecting muscle activity to control a video game. Rather than getting an open source game, rewriting the code, building a cable to connect it to the computer, and etc I found an old think geek DIY pong game.
)
