Comments on: NTT’s eye-tracking system monitors pupil size, blinking http://www.pinktentacle.com/2007/03/ntts-eye-tracking-system-monitors-pupil-size-blinking/ Fri, 05 Sep 2008 18:13:38 +0000 http://wordpress.org/?v=2.2 By: Sven Laqua http://www.pinktentacle.com/2007/03/ntts-eye-tracking-system-monitors-pupil-size-blinking/#comment-29328 Sven Laqua Fri, 16 Mar 2007 09:15:05 +0000 http://www.pinktentacle.com/2007/03/ntts-eye-tracking-system-monitors-pupil-size-blinking/#comment-29328 This sounds very nice, but it is still "light years" from realistic commercial use. I work quite a lot with eyetrackers and there are several that support these measures (pupil size, blinking, etc.) but the big problem is the restricted setup they will work with only. The system we use (Tobii) is the most advanced and flexible (for the user) that I have worked with, and market leader in the field. But still, there is no way to track people if they move their head more than about 10 cm in each direction - no go back to your sofa, and think about how you lounge in front of a TV. You might be able to connect some eyetracker with automated head-movement control, but then this will require mechanically moving parts and skew the accuracy of the data.... Sounds all very nice, but still way unfeasible in my opinion. This sounds very nice, but it is still “light years” from realistic commercial use. I work quite a lot with eyetrackers and there are several that support these measures (pupil size, blinking, etc.) but the big problem is the restricted setup they will work with only.
The system we use (Tobii) is the most advanced and flexible (for the user) that I have worked with, and market leader in the field. But still, there is no way to track people if they move their head more than about 10 cm in each direction - no go back to your sofa, and think about how you lounge in front of a TV.
You might be able to connect some eyetracker with automated head-movement control, but then this will require mechanically moving parts and skew the accuracy of the data….

Sounds all very nice, but still way unfeasible in my opinion.

]]>