We saw Steve Mann at the AWE 2013 and it was a good talk but one thing he said bothered me (so much I wanted to write about it;) He said that the Google Glass is a generation 1 Augmented Reality HUD and his glasses are a generation 5 HUD. He said at 17.20 that the physical camera has to be p.o.e. so in front of the human eye to allow a correct overlay of the augmented scene. This is not correct if the HUD display does only display the virtual scene and not the recorded camera image itself!
You do not need the physical camera to be in front of your eye to augment the virtual scene correctly on the HUD. When positioning the virtual camera you just have to take this physical 3d difference of the camera position and the HUD position in account and place the virtual camera with the same offset when augmenting a scene.
Why I made the effort to write a blog post about it and why think to explain the importance of this is that it means the camera does not have to be in front of your eye. It doesn't even have to be mounted to your body as long as the system knows where in the virtual scene your view point is located. So see through displays with transparent screens are possible (well the fact that the eye can only focus the display or the real world behind it is sort of a problem but this is another thing ;). It also means that in environments where the user can be monitored by an external system the only thing the AR glasses have to do is display a live stream to the user and they need no attached camera at all. Maybe in the future this becomes helpful ;)