recognition using wireless signals, it enables a user to perform gestures not only in front

of a computer or a TV but anywhere in the home.

WiSee works by leveraging wireless signals like those from wifi enabled devices around

us. the signals reflected off from the human body can be analyzed to detect human motions,

however, the signal received is broadband, making it difficult to sense subtle changes.

our algorithms tranform this signal into a number of very narrowband peaks of 2Hz, allowing

Wisee to detect changes in the signal on the order of 10Hz.

For example, when a user moves their arm, it causes change of 10-20 hz in the signal,

which wisee can detect.

In a environment with multiple people, wisee can lock onto a single user using multiple

antennas.

In this prototype implementation, we can see signal change in realtime as the user moves

his hand.

By analyzing the variations of this signal over time, we can enable complex full body

gestures that go beyond just hand motion. In this example, the user performs punching

and arm swinging gestures, each of which produces a distinct pattern. By applying machine learning

to these patterns, WiSee can detect what gestures are performed.

This sensing capability extends to the entire environment and not just in proxmity to the

wisee sensing device.

In a home setting, we envision WiSee to be easy to install as simply plugging in a wireless

router since our software could be built into it. Any wireless devices in a home, like a

phone or a laptop could be used as a signal source.

WiSee could then enable a number of exciting applications. For instance, a user could control

the music with a simple gesture. Similarly other devices could be controlled across the

room. This can extend beyond home entertainment devices to home automation, for instance,

controlling heating or cooling.

Since WiSee does not require the user to carry or be in line of sight of a sensor, we belive

WiSee could enable truly ubiqutous gesture interaction in the home.