Tobii EyeX

[Boff'in]

We managed to acquire an Eye trackers for, well, goof around with and develop a errr ... "research project".

We're trying to find ways that Eye tracking could be used more "naturally".

It's too inaccurate for actual aiming at discrete small targets outside of the central viewing cone, we are yet to test scrolling round movement.

(without testing) it feels like Eye tracking might be a better suited to what trackIR does, but instead of moving your head (and keeping your eyes focused on the screen) you just look around your cockpit.

Any future plans by the dev team to include an EyeTracking interface (for like the 10 people who have it )

 

[carlisle]

Would an eye tracking device not interface with the computer in the exact same way as a head-tracking device? Both of them input a two-dimensional set of coordinates so I don't see why not.

 

[Boff'in]

That's my aim.


It's just you don't have to rotate your head whilst keeping your eyes fixed on the screen.
you just move your eyes.

We're debating on which system to implement for our err research purposes.

 

[carlisle]

That's my aim.


It's just you don't have to rotate your head whilst keeping your eyes fixed on the screen.
you just move your eyes.

We're debating on which system to implement for our err research purposes.

I will have to look up this E.R.R. Research you speak of... is it a commercial operation or university funded? Surely 'err research' is code for looking at a certain kind of questionable material which would not pass the word filter on this forum... and I dread to think what you're doing with an eye tracker in that respect.

 

[Boff'in]

I'm on a uni course with software develpment and we have a module about interaction design.

We needed a project to work with,
I hastily said eye tracking.

And now we have a eye tracker, and we have to invent a research project around it (outside of the obviously boring studies of let's analyse where people look).

Within five minutes of getting the device set up, you realise accurate possitioning is a no-go, so pure mouse replacement is bust - straight off.
Not due to the hard-ware, but the wet-ware.
You only have a tiny window of vission of "decent resolution" - like a 3 cm diameter.
The rest of your vision is low-res and blurry.

You're eyes,both of them move multiple times (and in different directions) every fraction of a second to scan the area around.

And the brain is constantly stitching a mosiac of images together and pre-aggreed upon "pre-conceptions" to fill in the blanks,
and that puts the processing and filtering "lag time" down to like .1 of a second between you seeing something and you *cough* believing you experienced that thing.
You'll even see stuff which isn't there and ignore stuff which isn't suppose to be there - as per internal rules.

So yeah moving a mouse is bad, so fps, rts and "point and click" games aren't going to benefit unless your hot spots are huge.

So of course I'm into Elite so I thought a replacement for head-tracking.

 

[Tryst]

What distance does that 3cm visual cone represent? 3Cm over 10Cm means significantly more over 1 Metre. Can you not refine that to it's centre to create a point of reference for a mouse cursor at a given range from a screen, (say 1 metre)? Also adjust so the visual movements that scan the area could be eliminated by ignoring movements of less than a given amount, those movements are so tiny and quick, they are really no more than vibrations. Of course,this depends on how fast your tracking processor responds to any movement and translates it. Might seems a bit silly suggestion but maybe a slightly lower resolution optical pickup may help eliminate that "vibration".

Whoever uses it would have to "train" it in a similar way to the way we train voice recognition because everyone is slightly different in how their eyes move. Try several people finding a point on the screen as it disappears and reappears elsewhere, you might be surprised at how different the results are and it will also help you understand what obstacles you may encounter when it comes to different users.

It was a project a few years back by someone, (I believe Cal Tech but may be wrong as it was a fair number of years ago), but the processors were nowhere near fast enough back then, also the optical systems used to pick up the eye movements were nowhere near high enough resolution so the project was abandoned. I got interested in it and read about it and was quite upset when they abandoned the idea. The eye "vibration" was picked up by them with eye examination equipment used by opticians but their optical pickups didn't have the resolution to detect it so well, that may possibly work in your favour.

I'd love to see someone actually make a working version and I wish you all the best with it. I'd be one of those in the front of the queue to buy one if you manage to get it working and marketed.