Proof of concept demonstrates gesture recognition using WiFi signals

Computer scientists at the University of Washington have developed a technology that, with 94% efficiency, can recognize and distinguish nine separate gestures including pushing, punching, and full-body bowling.

This is significant for a couple important reasons.

First, unlike Xbox Kinect to which most people will immediately draw a parallel, WiFi signals are not limited by line of sight and can travel through walls. A common misconception is that signal cannot travel through concrete; in fact, it’s the metal rebar that degrades the signal. The device can even track multiple individuals discretely using multiple antennae.

And second, the technological hurdle overcome in sensing a change in signal of this magnitude is significant – only a few hertz when the signal source is broadcast at 2.4 or 5 gigahertz.

Applications include activations in which visual displays respond to crowd numbers or crowd movement, or smart-home applications which turn appliances, lights, or heat on and off when you arrive and leave respectively, or when certain gestures are performed.