In the Iron Man movies, Tony Stark never has to worry about interacting with JARVIS, the artificial intelligence in his “smart home,” by making sure that he stays in proper view of a motion-tracking camera. He just walks anywhere he damn well pleases, barks orders and waves his hands, and JARVIS is somehow able to track him. Now a team of engineers at MIT has created a system called WiTrack which could offer the same kind of experience. It’s a 3-D motion-tracking system like Kinect, except it doesn’t use cameras and it can track you anywhere in a house, even through walls.
Other smart-home experiments, like Frog Design’s RoomE, rely on fussy camera-based motion tracking systems to enable functions like pointing your finger at a lamp from across the room to turn it on. A system like WiTrack uses radio signals “whose power is 100 times smaller than Wi-Fi and 1,000 times smaller than cellphone transmissions” to reflect off a moving human body in space and determine its position in 3-D space with an accuracy between 10 and 21 centimeters. Right now the system can only track one person at a time, but the proof of concept implies that JARVIS-style, whole-home motion tracking could really be doable without festooning every room with tons of sensors and infrared cameras. (That said, WiTrack is still an academic research project, not a consumer product yet, so its own form factor is still pretty janky–the MIT video only shows it for a second or two and it looks like a messy clump of boxes and wires. The researchers claim that “WiTrack can be incorporated into consumer electronics,” though.)
“We believe there are endless possibilities in a smart home environment,” says Fadel Adib, one of the makers of WiTrack (along with Zachary Kabelak, Dina Katabi, and Robert C. Miller). “For instance, it can know when a person leaves a room and enters a different room, in which case it would automatically turn off the lights of the room he is leaving, and turn on the lights of the room he is entering. In another application, it can sense when a person gets out of bed and automatically turn on the hot water in the shower.”
These applications don’t exist yet, but one very useful one does: fall detection. WiTrack can detect someone falling down with 96.9% accuracy, making it a potential boon for shoring up safety in households where elderly people are living independently. That’s not exactly Iron Man-level interactivity, but if WiTrack or something like it saves your grandma or grandpa from a potentially fatal fall, I bet you’d consider it a very smart home indeed.