MIT Students Create AI Tech That Can See Through Walls
A group of researchers and students at the Massachusetts Institute of Technology have developed a technology that can identify objects and follow them even as they’re obscured by other objects. In a sense, that technically means it can see through walls.
It doesn’t actually “see” through walls, of course, because that’s impossible. Instead, the system uses radar-like technology to track people as they move around. It’s called RF-Pose, and early tests show it’s able to determine whether someone is walking, sitting down, standing, and waving.
The developers saw a success rate of 83 percent when it came to identifying individuals from a known group, and because of this, they claim the technology could one day be useful for various implementations, including police work, rescue operations, and health care.
RF-Pose Explained: Can It See Through Walls?
RF-Pose was developed by MIT’s Computer Science and Artificial Intelligence Laboratory, or CSAIL, with a team led by computer scientist Dina Katabi. It uses AI to teach wireless devices how to sense people’s postures and movement, even from the other side of the wall. The main component of RF-Pose is a large radio transmitter, beaming waves that can pass through walls but are reflected by human bodies because of the amount of water they store. A neural network is used to analyze these radio signals, and then a stick figure is generated — it walks, stands still, sits, and flings its limbs around as the person moves.
One potential implementation for RF-Pose is being able to monitor diseases such as Parkinson’s, multiple sclerosis, and muscular dystrophy. The data gathered by RF-Pose can paint a picture of how the disease is progressing, which doctors can then take into account and promptly adjust medication if necessary. The developers are currently working with healthcare professionals to explore these possibilities, according to MIT News.
RF-Pose And Privacy Concerns
For all its promising potential, however, RF-Pose presents some privacy concerns.
“If a normal camera is recording me, it means I am able to see the camera, too,” says Ginés Hidalgo, a Carnegie Mellon University research associate. “If this camera can be hidden behind or even inside any object, I would never be able to know when I am being monitored.”
Katabi acknowledges such concerns, calling it a valid question particularly in the current climate of data privacy. She ensures that RF-Pose anonymizes and encrypts the data.
The next goal for the team is to develop a version of the technology that can be marketed commercially.