I suggest you a table mounted one. There are a lot of models that are using IR lights and it should fit for your idea. Moreover, it would be nice also to record a video of the user driving (so you could study how tired is he)

Dear Ulrich, I would also suggest wearable eye trackers. You can try Tobii or SMI but you will have to add an extra IR scene tracking camera (like Raimondas suggested). I know that SMI has SDK and with a little bit of extra coding you would be able to extend their eye tracking glasses with additional IR camera. If you are working under Linux I would suggest you to write and create your own eye tracker. I developed an eye tracking software running under Linux. Let me know if I can help somehow.

Tomasz could you explain us how did you implement the calibration phase? Thanks

Hi Ulrich!

I used the precursor model of the DIKABLIS essential glasses in 2007 for a night vision study (during nighttime) in cars. It worked very well. The Institute of Ergonomics (Munich) uses DIKABLIS essential glasses and a faceLAB system in the static driving sim (http://www.lfe.mw.tum.de/en/research/methods-and-lab-equipment/static-driving-simulator/). For many studies the lighting conditions are rather dark, comparable to nighttime. That also works well. I also used the DIKABLIS essential glasses in four different flight simulators without problems.

Counter question: do you think that some systems are not suitable for use in dark environments? If yes, why?

Some thoughts...:

- In a dark environment you might try to avoid too many cables – somebody might fall over these cables and hurt/destroy something.

- Can you operate all equipment especially recording laptops in the dark? Illuminated keyboard?

- Synchronize the eye-tracking with driving data (can bus)? Other data streams? Equipment for other data streams? Operation of this equipment in the dark? Need some spotlights?

- Do you have/need a calibration process: once before recording / several times (maybe necessary with head-mounted systems)? Do you need light for the calibration?

- Do you have/need a field/scene cam? What about the dynamic range of this cam? It’s very difficult to record well illuminated small areas like the instrumentation displays, center information displays, secondary task displays in a dark environment. Many cameras are limited in their dynamic range and you can only concentrate on either the dark or the bright areas.

- Need some IR lights only visible for the field/scene cam but invisible for your eyes?

- What about small and immediate adjustments of the eye-tracker hardware in the dark?

- What about automated AOI detection in a dark environment? DIKABLIS needs markers which can be a problem when it’s too dark. Thus I always find positions in brighter areas or use spotlights.  

Regards from Munich!

Dear Andreas, Thank you very much, Fantastic advice. Have a nice day, Ulrich