The future of TV seems to be a big thing just now. The BBC recently had this report) from a company based in Israel called XTR3D. (Unfortunately, their homepage uses Adobe Flash, for no obvious reason.) In the report, they describe their gesture-based technology for controlling TVs. I thought it was interesting since I was part of a group looking at the same topic in university last year, however while they have been looking at developing cheap technology similar to Microsoft's Kinect, we looked at usability, going through two stages: asking people to suggest gestures, then testing the "learnability" of the most popular gestures. The outcome showed a flaw with gestures, and Natural User Interfaces in general.
TV remotes are clearly labelled; any gesture recognition system obviously can't have lables so the gestures need to be very obvious to everyone. Last year, we found that some of the fundamental gestures are fairly straightforward. People might get confused between changing channel and changing volume, but they are both quite clear and very easy to learn. However there are far too many controls for a TV for all of them to be natural. Even if you look at the simplest remote, the Apple TV remote which consists of a directional pad and three buttons, not all of those controls are clearly relatable to gestures. For example, what is the simple gesture for "Menu" or "TV Guide"? Our experiments found lots of good suggestions, such as opening a book, swiping up from off the screen, or even clapping, but there was almost no consistency, and they were quite tricky to learn. The same issue arose with "Play" and "Select". If you can't learn how to use your TV very quickly then you will most likely return it!
TV will probably never be controlled simply by gestures, as who wants to play charades with their televison? Using voice control alone will have to wait until AI can genuinely understand what we mean. In the meantime, both will be the equivalent of keyboard shortcuts - useful, but not used or even known about by everybody.