Why not put the camera on the oculus itself?

At this point I understand we've virtually moved past the point of DK1, but there did happen to be this curious device that credits to Robert MacGregor I happened to see a while back.

http://ovrvision.com/

While I understand the degree of difficulty it would be for an AR solution for the tracking. Perhaps it's possible with the massive amount of funding for hardware now that a solution and price are viable. Curious if it would be possible for a user to set up a beginning orientation assuming we've all but eliminated drift at this point. Do a calibration with the geometry of the room by slowly moving their head and having good enough recognition, and then moving back to the starting position. Granted it would probably work much worse in practice than in theory, and latency needs be considered.

Perhaps use Valve's odd QR design to signify for the device what the wall is and hence help with some of the depth issues that might come up. For instance one QR for each wall to signify what the farthest distance is. Granted not entirely sure a consumer would be willing to go through such troubles.

Apologies if such an idea was already alluded to earlier as well. Or if this just sounds plain like dumb conjecture.

http://en.wikipedia.org/wiki/Li-Fi would this give the camera something to position its self with if mounted on a rift?

instead of adding a camera to a rift, add projector to oculus rift http://www.eonreality.com/eon-icube/ ?