I think Kinects/distance detectors are still very young technology. Imagine how far it could be pushed with just engineering. A very high resolution detector could identify facial expressions, smiles, grimaces, smirks, a whole gamut of emotional inputs. Push the resolution even higher and you could use iris movements to zoom in and out based on what the user is trying to look at.

Voice control was omitted from the video, I don't really understand that. I still think natural language is potentially the best mode of interaction and other modes would be secondary after that.