From: dstamp@watserv1.waterloo.edu (Dave Stampe-Psy+Eng)
Subject: Re: head tracking
Date: Wed, 27 Nov 1991 18:56:33 GMT
Message-ID: <1991Nov27.185633.16876@watserv1.waterloo.edu>
Organization: University of Waterloo


brucec@phoebus.labs.tek.com (Bruce Cohen) writes:

>In article <1991Nov25.211301.21688@milton.u.washington.edu> azuma@cs.unc.edu 
>(Ronald Azuma) writes:
>
>>         This system is an experimental one and has problems.  The biggest
>> is the combination of weight and lack of head rotation range.  The "cameras"
>> and lenses are too heavy, which limits us to four "cameras," which is not 
>> enough to let you tilt your head as much as you would like.  We need to make
>> the head-unit lighter so we can eventually put more "cameras" on the head 
>> that will provide the desired rotation range.
>
>Could you explain why you decided to put the LEDs in the room, and the
>cameras on the head?  I would have thought that putting the LEDs on a
>headband or some such would be easier on the user, and would allow
>better coverage.  For instance, bending over could be handled if there
>were two sets of LEDs, one pointed generally upward and one downward,
>with the sets spaced around the circumference of the head.  You could
>easily use 8 or 16 LEDs, so each would have to cover much less than an octant
>of (dare I say it?) headspace.

I would hazard it's because of the improvement in head angle resolution.
If your head rotates thru N degrees, the LED position in the camera's view 
changes by N degrees.  If you are looking at the relative position of 3 
LEDs on the head from a fixed-base camera, the resolution depends on
the distance and the spacing of the LEDs.  About a 5:1 difference, depending
on the camera's FOV.

I'm guessing that the LARGE number of LEDs on the ceiling attempts to place
a number (6 to 9) within the camera's narrow FOV.  That makes it easy to 
solve the perspective-transform matrices for position and angle.

I STILL like the idea of a standard head-mounted camera best, though.  
It's the easiest to use, it's off the shelf, and current models are
pretty light.  Or you can use a cheap vidicon model (Sony Watchcam)
which is nearly as light as the "finger" cameras.

And, since you need only 1/3 degree rotational accuracy, you can use a
large FOV (>100 degrees) which means you need only a few LEDs scattered
throughout the room.  Correct for the inevitable wide-angle lens distortion
mathematically.


--------------------------------------------------------------------------
| My life is Hardware,                    |                              | 
| my destiny is Software,                 |         Dave Stampe          |
| my CPU is Wetware...                    |                              | 
| Anybody got a SDB I can borrow?         | dstamp@watserv1.uwaterloo.ca |
__________________________________________________________________________