The way it works is that you have a camera that sees what you would normaly see, have the image be processed by a computer so as to extract features that the computer recognise. The computer then "augments" this image by adding some information (overlays) to the image, and displays this image in front of the eye.

You can do this with one eye, or, for normal people, with 2 eyes.

The setup looks like a VR920 on which you add one central camera, or one camera per eye (you can play with the separation to add depth of field, you can play with what the camera actually sees: visible light, infrared, ultraviolet, x ray, whatever ;-), you can add some lighting too)...

Now from the vuzix perspective, it is very possible to create a separate product that would integrate all of this, or they can see this as separate addons that VR920 customers would buy to accomodate for their specific AR needs.

http://fr.youtube.com/watch?v=ItOtTdhDoto