June 4, 2013 at 6:25 PM ET
Want to switch off the living room lights from bed, change channels while washing dishes, or turn the heat up from the couch? A team at the University of Washington has rigged a standard Wi-Fi home network to detect your movements anywhere in the home and convert them into commands to control connected devices.
Gesture recognition is the latest fad in games and tech, but even the newest systems require high-tech depth-sensing cameras or other special hardware. Microsoft's new Kinect, for instance, uses a photon-measuring method called "time of flight" sensing that was, until the Kinect was announced, limited to high-tech laboratories. And Kinect isn't small, either.
UW computer science students, led by assistant professor Shyam Gollakota, looked at the gesture-detection puzzle another way — specifically, how people affect the environment they're already in.
Our bodies distort the Wi-Fi signals we use to beam information to and from our laptops and phones. By watching those signals very closely, the team could determine not just what room you're in, but where you're standing and how you're moving your body. They call the system WiSee.
"By analyzing the variations of these signals over time, we can enable full-body gestures that go beyond simple hand motions," said Qifan Pu, a visiting student and one of the team at UW, in a video outlining the work.
That's no easy task: the "doppler effect" that our bodies have on the wavelength and path of the Wi-Fi signals is miniscule, meaning reliable measurement with consumer-grade hardware is difficult. But the WiSee team's expertise worked it out.
Once the sensing process was rigged up, the group combined the gesture recognition with store-bought home automation devices that wirelessly control lights, media players, thermostats, etc. Soon, they were using WiSee to perform simple tasks like playing a song or changing channels.
The system is also capable of tracking people as they wander through rooms or out of the house, turning off lights or adjusting music volume depending on their location.
The team put together a prototype piece of hardware to demonstrate WiSee, but any modern Wi-Fi router should do the trick, too, with a bit of custom software. With no special devices to buy, this could be the cheapest gesture-recognition tech yet.
Don't worry about anyone installing it surreptitiously on your router, though: It takes a bit of expertise and some specific "training" of the software before it can recognize anything at all, much less specific gestures or locations.
PhD student Sidhant Gupta and assistant professor Shwetak Patel, also on the project, have worked with Microsoft Research on similar body-tracking systems, but using soundwaves or radiation from electrical wires as the medium.
WiSee is currently in the proof-of-concept stage, but the creators hope to present it at the International Conference on Mobile Computing and Networking in Miami later this year.
More information, including a technical description of the system, can be found at the project's website.
Devin Coldewey is a contributing writer for NBC News Digital. His personal website is coldewey.cc.