Excuse me if I once again bring a more lunatic note into your refreshingly technical conversation. You do extremely cool work guys, I'm just itching to join the wearable craze myself. (But for time/$$$'s) Due to the lack of responses whether the borg-like PPP/IrDA linkup between two wearables has been tried I assume the negative. I have been told the IrDA specs suck, and the protocol stack is difficult to implement. Oh well. But imagine talking over a few 100 m with a line-of-sight (in the verbatim sense) laser link attached to your display. If you have a CCD webcam, you can use NIR as a light source/visual feedback for homing onto another person's photoreceptacle. NIR LED cluster lamps are good for the reeeal limelights (juice! juice!). I'd like to bounce off you a couple of other pretty obvious things. You probably know about IBMs PAN (Personal Area Network) thing, where you use galvanic potentials over the 'handshake' protocol (handshake like physically shake hands with a person) to exchange a few 100 Bytes of personal data, like ID/email address, time stamp, GPS coordinates, PGP public key, whatever. This should be pretty easy to implement DIY. Anyone tried this? (Btw, are you aware that you are defining future de facto standards? Investing some wetware CPU hours into specs/coordination among developers might avoid lots of future oopslas. But you know this already). Has anyone implemented impromptu EM microcellular networking by having just a bunch of persons in the area (i.e. using your wearable as router)? Ok, I admit this is pretty advanced. Anybody tried Speak Freely over a headset/cellular modem/IrDa link? Btw, beta conferencing/sound reflectors has been made part of Speak Freely for Unix a couple of days ago. I understand there are no immersive displays out there yet. Still, since some of you are using wearcams, has overlay over the video been tried? Piping in video from a backwards-pointing CCD camera? Voice recognition has been mentioned. Sadly, last time I looked voice recognition support for Linux was still very raw. Nevertheless the ability to key certain things like video/audio logging or snapshot a datestamp/GPS position fix via a voice command would be very good to have. Juice is a grave problem, since it translates in weight and operation hours. You probably know about fuel cells. There are now RT alcohol (methanol) metabolizing cells in existance, albeit the materials are pretty expensive (polymer membrane, platinized carbon cloth), and you'd have to DIY a fuel cell stack with enough power. Still, this is very interesting future option. On a much more fringeoid note, MEMS microturbine generator arrays could become available in a decade, or so, so you could just refuel at a gas station ;) Several of you probably toyed with GPS. I used a Garmin receiver, because their OEM developer stuff was relatively affordable. Still, boards based on Sirf's chipset should be vastly more useful in typical urban environments, any experiences with these? Btw, I think I saw some rudimentary GIS map display map with GPS support (position blip overlay onto map, so they must have a NMEA parser) in a Linux contribution directory somewhere (sunsite?). Garmin uses a neat visual guidance aid (a stylized highway to the target), this would be also very good to have as a video overlay. Do you also display date/time in the visual field somewhere? Other stats, e.g. temperature? What else? Analog displays or digital? GPS doesn't give you realtime heading info. Magnetic sensors are out there, their input would be nicely complementary to the GPS stuff. Of course there are lots of magnetic anomalies in urban environments... Which brings me to MEMS accelerometers. Since some of them offer few ug-resolution, while being remarkably shockproof, having inertial position tracking with 3 orthogonally aligned sensors appears viable. This is also nicely complementary to Hall sensor/GPS data. Of course they are not easily obtainable, and writing driver stuff would be hideously difficult. But not for MIT guys, these can walk on water ;) There are eye trackers out there. You could either evaluate sideways-reflected IR light or brute-force DSP the video feed from an RT CCD eye image. The eye being a dipole inducing potentials in nearby tissue can also be evaluated. (Gives you information about how much light the retina receives, too). Which brings me to Residual light amplification. A russian fleamarket device the size of your thumb was very useful, and clearly saw a NIR remote control in a dark cellar. I could even use it for a light source. Since you transatlantics have access to latest generation multichannel plate devices, this oughta be a home run. Using biofeedback techiques, you can extract few bit/s from OPAMPd scalp/brow myopotentials (AgCl electrodes in a headband, or geodetic electrode array in a HMD mount helmet -- bald gentlemen are obviously preferable), or patternmatch FFTeed EEG. The latter is really tricky, but amateurs have purported it to work. If you train it real good you should be able to intuitively steer a pointer (nudging by increments), trigger some devices, and generally mess up your EEG pattern ;) Apropos feedback, anybody monitoring bodily functions? I toyed with a salvaged PulseWatch once, but it did not live very long. It uses weak RF link for about a 1 m range. Embedded PCs are not very good embeddeds. Using true embeddeds with CAN support/existing protocol stack can make things both speedier and cheaper. I think some of the CAN-capable controllers do have flash. Which brings me to another one: anybody using Forth in a wearable? Yes, I like rhetoric questions. Ah, another thing I almost forgot. In the context of augmented reality a 3d pointer (magic wand) would be great. Easiest would seem to be a doppler/TOF tracker, the more so because I believe an affordable CRT-mounted tracker for gaming purposes does exist. I surely forgot several things in the rant, but this is probably too much already. Comments welcome. ciao, 'gene
From Wear-Hard Mailing list Archive (WH)
Maintained by R. Paul McCarty
Archive created with babymail