OCZ Neural Impulse Actuator (NIA)

Triathlon is inactive atm.

There used to be a long thread on the old OCZ NIA forum, but that has been closed, it appears.

http://sourceforge.net/apps/mediawiki/triathlon has some infos, but nothing recent.

Triathlon was my first project in Python ever, so the code is as ugly as it gets. And the use of wx for GUIs has been a bad choice as well (lots of cross-platform issues).

I think, I will rewrite a lot of it sooner or later.

Overall the approach is not as silly as one might think and the crude evolutionary neural-network breeder can be useful far beyond the scope of Triathlon.

There is some uncertainty around the future of the NIA ... ( http://www.brightsideofnews.com/news/2011/4/12/exclusive-future-of-brain-computer-interface-endangered-by-a-ponzi-scheme-investor.aspx ) ... so that's a bummer.

On the other side, NIA and Rift headgear could be merged really nicely, it appears.

So right now, Triathlon is one of those inactive projects that I am still always thinking about.

By now, I have realized that future Brain-Computer-Interface software will follow a very different paradigm than simply "thinking commands". What we truly want is a BCI-influenced autopilot. So the interpretation of EEG-signals should take more data into account. a.) more sensor-data and b.) game-engine data or video or stuff like that.

The final result would be a BCI-influenced (game-)bot thingy.

Is the Emotiv Insight a legit product?

Edit: (for grammar fixes so that point comes off more clealy)

I looked into it a bit more, and it's obvious that such products as the emotive insight are good things, steps in the right direction, and worth looking into for research and development. I'd suggest investing in one to tell the developer to keep making more. On the other hand, however, I believe the Emotiv Insight is pretty gimmicky, especially as technology that's going to be available to the public. If you were going to throw down the money for the kickstarter and want a full-fledged input device, I would suggest purchasing an Emotiv EPOC because of the steep difference in ability.

Then, if you wanted to be a games producer (not necessarily a programmer) using the EPOC you don't have to flop out the whole $500 for the devkit (expensive, but that's to be expected). If you're not against using a game engine, the site has an $80 Unity license for using the Epoc in that game system. One limiting thing i've noted, however, is that they seem to want to use their store to allow for purchases, similar in model to the announced oculus store. I have not read if they want your app to be exclusive or not to their store, however.

Epoc w/ unity:
http://emotiv.com/store/apps/applications/131/4021

Makes me want an epoc... Which makes me sad my current computer is so... un-good.

What is not demonstrated by the insight, is the use of multiple commands by one user. It states that one can set it to recognize 4 separate states of thought to perform actions, when there are 6 available... I did put money into the kickstarter, mind you- however, I am debating on changing to order an epoc instead.

My verdict: The insight is useful for technology demonstration, not necessarily household use, besides people such as paraplegics, etc. Its primary features will (I imagine) be something like the facial recognition and attention level. I would, however, like to have both the Epoc as well as the Insight to compare the two.

Also, seeing as to how the kickstarter reached its stretch goals, they will be including 6 new ways of getting data - via the 3 axis of head position/movement, similar to the rift's capabilities - significantly expanding the abilities of the Insight. I'm not sure what to do with such a tool, but I am excited to see these types of things for public use.

That said, if you want to invest the product and show it off to others- go for it.

I'm making the call that the product will be less than it proposes to be, especially with the distinct lack of... how do you put it? Discussion on the product's limitations.

As a note, i did pledge to that kickstarter... Can't decide if I should go for the devkit at 429, OR the research kit. Both aren't an option, or I would consider the 529. I don't think i'll use the research software, as much, so I think I will eventually go for the dev kit version instead. :P

On the other hand, I am interested in the abilities of gathering the raw data... GAH. :(

"dwood15" wrote:
I would suggest purchasing an Emotiv EPOC because of the steep difference in ability.

I also backed the Emotiv Insight, but your comments have me a bit confused. From what i read elsewhere (on neurogadget) i thought the insight was meant to be a successor of the EPOC, not a downgrade though i can certainly see the EPOC has more sensors (their price also being quite similar, $229 kickstarter vs. $299 for the EPOC).

Personally i am going for the "consumer" version just to see how well it works, if i am impressed, i am hoping it will be possible to buy the developer/research SDK later on ( i guess the hardware is identical? ). I am hoping that at the very least i might be able to use it as a relaxation/mediation help, but issuing simple mental commands would also be fun.

Anyway, the insight has dry sensors, messing with the saline solution for the EPOC seems a bit of a hassle.

Can anyone provide any feedback on the performance of the NIA's "Retinal Movement" reading performance? I've been looking around for answers on various forums and sites, but the point tends to get glossed over or ignored. Me and my group have actually found a use for such eye tracking in our projects, but would like to know if it could be possible to achieve while wearing the Oculus Rift. We don't need particularly high precision due to the way we're designing things as any form of eye tracking solution would be great at this point, though we'd be delighted if the NIA could be used like a rough mouse cursor. If anyone knows any other potential solutions, do mention them, though perhaps that would be best for the eye tracking forum thread.

eye focus tracking... eyelines... the ever and still elusive prize.

I wouldn't hold your breath too hard on this particular puppy guys.

IMO this is still very carefully guarded military application technology that most likely won't trickle down until such time as a newer better targeting system is perfected, AND a universal remote countermeasure for the existing application is perfected.

think about it. and the damage one could do with it, if one was so inclined.

to expand on that last thought and offer some constructive goals one could pursue...

context and action and even "guess the word one is thinking" histogram generatikon based prediction, relying on eye movement, microtwitches and tracking alone IS doable, has been done for some time now, and is out there, commercially available.

what you're looking for is facial electromyography. one therapeutic technique which uses something similar is called EMDR... eye movement something reconditioning... they use it to offset the neural cascade which precedes intense post traumatic stress disorder anxiety attacks or flashbacks.

facial electromyography is damn effective. really effective, but it has some serious drawbacks and IMO, if you want to use it on its own for consumer entertainment purposes, it's going to be a dead end which isn't worth your time to pursue.

it's so good that it can predict what words you're thinking about, by taking good educated guesses at each consecutive phoneme till the phonemes become a word, and the words become a sentence, and so on.

it works. like it's reading your mind, kind of.

i say kind of because it is fatally flawed.

if you have a big chit eating grin on your face, as most gamers using this headset are likely to have, it's next to useless... LOADING... PLEASE WAIT.... LOADING... PLEASE WAIT... LOADING... PLEASE WAIT....

If you're laughing hysterically to yourself, even just on the inside, non-verbally, same problem.

blank 1000 miles away vietnam veteran stare toward the horizon... same problem.

but... that said... if you want to use it as ONE approach to creating a thought predictive histogram UI core, driven by a software based Autonomous Digital Intelligence avatar or agent on the client side, then you should most definitely pursue this, because it really, really, really works, until you grin like a cheshire cat.

peace

to be honest and cut to the chase...

I think the UI approach here needs to be a day zero ground zero approach, like the oculus itself.

and yeah, ok, it's not a NEW new concept or design. but what's new is that consumer tech has caught up with high end pie in the sky eye in the sky I spy with my little eye guy cold war thingies.

NSA appears to have tricked down the satellite imaging tech. Smith Micro's Photo Zoom appears to do what was whispered about in dark corners, with satellite images that could read a newspaper on the deck of a ship.

other stuff is appearing too, day by day.

and with the Unreal engine's patent renewal expired late last year, meaning the DARPA and NASA/JPL funded physics engine also becoming public domain, and Linden Labs Second Life already incorporating most of it and moving the render server side, and more patent expirations coming at us fast and furious day by day from here on out, for the remainder of humanity's future, it's an indie developer's Wild Wild West Wet Dream out here at the moment.

Space Age tools.

Literally.

You know what I think?

I think it serves very little purpose to re-do what's already been done and is being done in evolving forward motion by Linden Research.

I think a collaboration between Oculus VR and Linden Research is a perfect pairing.

And for those of you scoffing at this... I'm guessing that you haven't popped your head into Second Life recently and taken a look at the debug settings.

Prepare to be eye popping wowza'ed.

You can client side define the scale of texture uploads now. And they can handle AO, bump, spec, displacement, glow, diffusion, shaders, reflection, the whole kit and kaboodle.

AND... you can composite layer to your hearts content and put it all one one texture, and their render engine's texture analysis will tear it all apart and find what's there and apply it in what seems to be the most logical and optimal fashion.

But Linden Research, which now has the render engine tools to be Virtual Hollywood, for real... I'm talking realtime render broadcast quality here... also has a major gap that needs filling.

They need an interface that will help them move from static 30 and 60 second loops to freeform off the rails end user control, so that my actors can act, and not simply be underutilized stand-ins.

The value of the physical actor... Chevy Chase, Steve Martin, Jim Carey, Charles Chaplin, John Cleese... this art will be lost to us without someone giving the community a freeform Pixel Bumping Kung Fu... Pi Bu Kung Fu for the 21st century.

And THAT is where oculus and this community step in, IMO.

And I'm going to lay out some broad strokes ideas for people to pursue over the next couple of days.

So stay tuned.

Peace

"LokiMortmagus" wrote:
to expand on that last thought and offer some constructive goals one could pursue...

context and action and even "guess the word one is thinking" histogram generatikon based prediction, relying on eye movement, microtwitches and tracking alone IS doable, has been done for some time now, and is out there, commercially available.

what you're looking for is facial electromyography. one therapeutic technique which uses something similar is called EMDR... eye movement something reconditioning... they use it to offset the neural cascade which precedes intense post traumatic stress disorder anxiety attacks or flashbacks.

facial electromyography is damn effective. really effective, but it has some serious drawbacks and IMO, if you want to use it on its own for consumer entertainment purposes, it's going to be a dead end which isn't worth your time to pursue.

it's so good that it can predict what words you're thinking about, by taking good educated guesses at each consecutive phoneme till the phonemes become a word, and the words become a sentence, and so on.

it works. like it's reading your mind, kind of.

i say kind of because it is fatally flawed.

if you have a big chit eating grin on your face, as most gamers using this headset are likely to have, it's next to useless... LOADING... PLEASE WAIT.... LOADING... PLEASE WAIT... LOADING... PLEASE WAIT....

If you're laughing hysterically to yourself, even just on the inside, non-verbally, same problem.

blank 1000 miles away vietnam veteran stare toward the horizon... same problem.

but... that said... if you want to use it as ONE approach to creating a thought predictive histogram UI core, driven by a software based Autonomous Digital Intelligence avatar or agent on the client side, then you should most definitely pursue this, because it really, really, really works, until you grin like a cheshire cat.

peace

Seems like that could be an issue as you've stated, though anything at this point could be helpful at this point in time in the situation I'm looking at (it's mainly for aesthetics, though it the use of eye tracking for more game play centered goals is still on the table, which isn't visible with the Rift.

Back to the point of the NIA, has anyone tried using it for the purpose of eye tracking or at the very least for some rudimentary, look left and right controls. Since it seems to only work for lateral movement I've figured that using the NIA's muscle/tension sensors to support this lateral movement could create a somewhat basic, but functional "gaze" tracking system that could be used in targeting for Rift applications. If a facial expression of joy or the like can reduce the accuracy that much, it may not be particularly viable and other systems by have to be explored. Unfortunately, camera based eye tracking my not be implementable at this time without significant jury rigging techniques and unless Oculus does choose to implement it in the consumer version, the concept may be left to other EOG systems or to the Emotiv Insight (If that's possible, I haven't been able to conclude if the SMI connection implies that or if it's just for providing emotion feedback to the data provided by SMI's own products, which wouldn't be useful for gaze tracking; though re-rigging the though detection used to move the box could be useful as a support for this or at the very least a temporary use (planned to be use for another purpose at this point, separating it from eye tracking would be optimal)). I suppose I'll get back to doing the research myself, though I'd prefer not to have to buy the device if the purpose I have in mind can be verified as not working to begin with. Hit and miss is a very inefficient way of going about things.

OCZ NIA will not be suitable for eye-tracking.
You can very reliably detect eye-movement ( http://en.wikipedia.org/wiki/Electrooculography ), but you wont be able to figure out the direction your retina is pointing, from that data.

So:

Eye-tracking - no
REM-detection - yes

In the official NIA jargon it is called "Glance". And they try to guess a direction for that "Glance", but it doesn't work reliably.

NIA with REM-detection has been part of experiments around lucid dreaming in the past.