A prototype of a wearable device can sense what appliance you’re using. Engineers at the University of Washington developed MagnifiSense, a wrist-worn magnetic sensing system that tracks your interaction with specific devices, such as a microwave or hair dryer. Based on which device is detected, the system infers what activity you’re performing: Turning on a stove implies that you’re cooking, for example.
Tracking an individual’s daily activity could help monitor and perhaps reduce a user’s energy footprint, or it could feed data to smart home applications and provide safety alerts for the elderly. The Washington team described MagnifiSense and its potential uses in research [pdf] presented at the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing in Japan last week.
MagnifiSense works because each appliance generates a distinct electromagnetic radiation pattern. MagnifiSense uses off-the-shelf magneto-inductive sensors to capture a wide spectrum of frequencies near the user. This allows the wearable device to identify the radiation of the particular components—motors, rectifiers, and various modulators—that make up the pattern. Using signal processing and machine learning techniques, the system can use the combination of components to distinguish one device from another.
Edward Wang, lead researcher and a PhD student in electrical engineering at the University of Washington gave a hairdryer as an example:
The frequency component of a hairdryer is that there’s a motor that spins, so there’s going to be some changing frequency that has to do with the motor’s speed… There’s also the power that it draws, which is 60 Hz in America. The 60 Hz component can be seen in our signal. So, if it doesn’t have a 60 Hz component signal, then it’s not plugged into the wall.
These type of characteristics, also known as domain knowledge, are gathered into a feature set, which is similar to a template, says Wang. After hundreds of different hairdryer templates are fed into the system, it learns to identify the behavior of a hairdryer. Then, when the template of an unknown device is fed into the system, it compares it against existing templates to determine a match.
The team studied MagnifiSense’s performance in 16 homes and on 12 commonly used appliances in the kitchen, living room, and bathroom. It also studied the user’s interaction with various devices. In a 24-hour period, MagnifiSense successfully identified 25 of the 29 interactions.
Although this technology seems promising, researchers still need to work out a few kinks. People tend to interact with multiple electronics simultaneously. But, the current prototype can’t detect when multiple electronics are used concurrently.
“Due to the nature of the signal, they add linearly,” Wang says. “The sensor sees A plus B plus C.” This means that if you turn on the stove and also use the blender, the system detects the appliance closest to you- not both. He also says they’re trying to miniaturize the wearable device.
Theresa Chong is a video host and multimedia technology journalist based in Palo Alto, Calif. As on-camera talent, she has performed science experiments for “Discovery News,” explained how virtual reality works for USA Today, and interviewed Adam Savage for IEEE Spectrum. She has written about wearables for Scientific American and travel tech for Architectural Digest. With a DSLR, GoPro, and green screen by her side, she has produced digital videos of robots, driverless cars, and 3D printing. She earned a master’s degree from Northwestern University’s Medill School of Journalism, and in a prior life she worked as a civil engineer.