Would there be any interest in building a mems gyro/accelerometer based head
tracker? The nice thing about a vision based head tracker is that it's
passive an you don't need to attach any equipment to your head or your cap.
But obviously there are limits to what you can do (well and there are
limits of course with any approach.)
An IMU based head tracker would offer a lot more range and flexibility, but
you would need to attach it to a cap and wear the cap ... and have some sort
of umbilical cord connected to your computer.
For instance, look at this interesting product. For $125 you get 3 axis
gyro, 3 axis accelerometer, 3 axis magnetometer in a package that includes
"open source" firmware to implement a simple AHRS filter and it talks to
your computer via a serial connection.
Something like this could potentially capture the roll, tilt, and yaw of
your head as well as relative motion in 3d. It seems like a person could
design a simple filter that assumes your head motion will always be around
some zero reference point and can use that assumption to "correct" for
drift.
http://www.sparkfun.com/commerce/product_info.php?products_id=9623
Now here's my other question. Would any kind of head tracker be all that
effective when you are looking at the virtual world through a monitor?
Maybe? But it seems to me that if you combined a simple head tracker with
some sort of head mounted displays, then you would really get into some
interesting territory ... especially in combination with the amazing 3d
cockpits that are included in many flightgear aircraft. You could look
around the cockpit, you could stretch up and side to side to look over or
around the nose of the aircraft or look around the "A" pillar (that's a car
term, so whatever it's called in the aviation world.) If you could track
your head motion accurately enough to not make yourself sick, it could be a
pretty incredible system.
Regards,
Curt.
On Sun, Feb 28, 2010 at 2:11 PM, Melchior FRANZ wrote:
> Head trackers determine the position and orientation of the computer user's
> head, and use this information to modify the view in the artificial 3D
> world.
> Usually, this requires a sensor (e.g. a webcam) and special markers mounted
> on a helmet or cap. This can be colored/reflective points or even (IR)LEDs.
>
> There's a (F/OSS) implementation of a head-tracker that only requires a
> webcam, without the head gear: EHCI[0] (based on OpenCV[1]). It works with
> a face detection algorithm. See the following link for a demo. It shows the
> analyzed face features on the left side, and an overlaid 3D head on the
> right side.
>
> http://www.youtube.com/watch?v=BovphSjw_tI
>
> Unfortunately, it doesn't work as well as the film implies. The guy took
> care of not turning the head too far and keeping the face rectangle nicely
> on screen. If one doesn't do that, then the orientation quickly drifts off,
> and one has to reset the tracker.
>
> Anyway, in the hope to see further EHCI improvements in the future, I've
> written an fgfs addon that uses the tracker information to control the
> pilot's view. It's a separate application that communicates via UDP socket
> with fgfs, and there are no changes to fgfs (2.0!) necessary. See the
> README
> for further info:
>
> $ git clone git://
> gitorious.org/fg-ehci-headtracker/fg-ehci-headtracker.git
>
> Maybe someone is interested in this *and* can perhaps even send me patches
> for improvements/fixes. :-)
>
> Note that the EHCI library is a bit strange (to put it politely), which
> explains some of the fg-tracker weirdness. Also note, that you have to
> use it in not too dark environment (or maybe use an IR spot, haven't tried
> that yet). And I assume you should have a dual core/cpu machine at least.
> The tracker uses quite some cpu power for image processing, so it shouldn't
> run on the same cpu/core as fgfs.
>
> m.
>
>
> PS: don't expect too much!
>
> [0] http://sourceforge.net/projects/opencvlibrary/
> [1] http://code.google.com/p/ehci/
>
>
> ------------------------------------------------------------------------------
> Download Intel® Parallel Studio Eval
> Try the new software tools for yourself. Speed compiling, find bugs
> proactively, and fine-tune applications for parallel performance.
> See why Intel Parallel Studio got high marks during beta.
> http://p.sf.net/sfu/intel-sw-dev
> _______________________________________________
> Flightgear-users mailing list
> Flightgear-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/flightgear-users
>
--
Curtis Olson: http://baron.flightgear.org/~curt/
------------------------------------------------------------------------------
Download Intel® Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
_______________________________________________
Flightgear-users mailing list
Flightgear-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-users
