From zhai@almaden.ibm.com Fri Jun 11 20:08:04 1999 Received: from burdell.cc.gatech.edu (root@burdell.cc.gatech.edu [130.207.3.207]) by lennon.cc.gatech.edu (8.9.1/8.9.1) with ESMTP id UAA23935 for ; Fri, 11 Jun 1999 20:08:03 -0400 (EDT) From: zhai@almaden.ibm.com Received: from asbestos.hitl.washington.edu (hitl-new.hitl.washington.edu [128.95.73.60]) by burdell.cc.gatech.edu (8.9.1/8.9.1) with ESMTP id UAA08101; Fri, 11 Jun 1999 20:07:45 -0400 (EDT) Received: from e4.ny.us.ibm.com (e4.ny.us.ibm.com [32.97.182.104]) by asbestos.hitl.washington.edu (8.9.3/8.9.3) with ESMTP id RAA00386 for <3d-ui@hitl.washington.edu>; Fri, 11 Jun 1999 17:07:11 -0700 (PDT) Received: from westrelay02.boulder.ibm.com (westrelay02.boulder.ibm.com [9.99.132.205]) by ny.us.ibm.com. (8.9.3/8.9.3) with ESMTP id OAA229912 for <3d-ui@hitl.washington.edu>; Fri, 11 Jun 1999 14:55:15 -0400 Received: from d53mta03h.boulder.ibm.com (d53mta03h.boulder.ibm.com [9.99.142.3]) by westrelay02.boulder.ibm.com (8.8.8m2/NCO v2.01) with SMTP id MAA68388 for <3d-ui@hitl.washington.edu>; Fri, 11 Jun 1999 12:55:19 -0600 Received: by d53mta03h.boulder.ibm.com(Lotus SMTP MTA v4.6.4 (830.2 3-23-1999)) id 8725678D.0067EF76 ; Fri, 11 Jun 1999 12:55:15 -0600 X-Lotus-FromDomain: IBMUS To: Matt Conway cc: 3d-ui@hitl.washington.edu Message-ID: <8725678D.0067E8B1.00@d53mta03h.boulder.ibm.com> Date: Fri, 11 Jun 1999 11:54:55 -0700 Subject: Comments on Eye Tracking Technologies Mime-Version: 1.0 Content-type: text/plain; charset=us-ascii Content-Disposition: inline Status: RO Matt, Eye tracking has become increasingly popular in recent years, both as a UI analysis tool and as an interaction technology. So I think it is worthwhile to share our experience with the group. We did market research on eye-tracking technologies about two years ago. In the end, we choose ASL, because it seemed to be the largest vendor and it promised high percentage of users who can be reliably tracked. It also promised a cubic foot of head movement. We were very disappointed. First, it could not easily track me and many of my colleagues. It may work better if you do a lot of tweaking (threshold, lighting, camera position, screen brightness etc etc) but the same adjustments may not work the next time you come back to the tracker. It does work reasonably well if you have "good" subjects: people with no glasses, with large and bright pupil etc. Younger people tend to be better subjects too. Second, the 1 cubic foot of head movement is simply not true, although there is a slow servo on the camera. If you want reliable data, the subject has to stay steady. Having said that, I am not suggesting ASL is not among the best. I do not believe any one has a truly satisfactory eye-tracker yet. We eventually developed our own eye-tracker. The key idea that made our eye-tracker different from the commercial ones is that we used a dual infra red illumination scheme, so both the dark pupil and bright pupil (like red-eye in photos) are detected. It worked a lot better (still far from ideal). I hope the commercial eye-tracking companies will follow the idea soon ( In fact we re-invented the idea, it has been around, but not very well known). For a brief description of the IBM Almaden eye tracker, check Zhai, S. Morimoto, C., Ihde, S. "Manual And Gaze Input Cascaded (MAGIC) Pointing", Proc. CHI'99. Shumin ___________________________________________ Shumin Zhai, IBM Almaden Research Center, Tel: (408)927-1112, Fax (408)927-4366, http://www.almaden.ibm.com/cs/people/zhai