Carnegie Mellon University researchers have developed a tracking device out of a common household wifi appliance. They were able to detect and track the three-dimensional shape and motion of people within a room by using the radio signals provided by WiFi networks. They did this remarkably well without the use of cameras or expensive LiDAR technology.
According to the study’s authors:
We argue that WiFi signals can, in some situations, replace RGB images as a universal tool for human sense. WiFi-based solutions used for indoor monitoring are mostly unaffected by ambient light levels and partial or complete obstructions. They also prevent invasions of personal privacy and the necessary equipment is affordable.
The researchers used DensePose, a method developed jointly by a team from Facebook’s artificial intelligence lab and a company located in London, for assigning each pixel on the surface of a human body in a picture.
DensePose’s strength is that it can identify more than two dozen important anatomical sites and zones, including joints and different body components like the limbs, head, and chest. Because of this, the device can also detect a person’s posture with great precision.
The WiFi technology is very impressive
The team was able to accurately map the phase and amplitude of WiFi signals emitted and received by routers to particular places on human bodies by combining this massive amount of data with a deep neural network.
Three inexpensive WiFi routers (each costing the researchers a total of $30) and three matching receivers (each costing $30) were used to illustrate the concept, with the signals being redirected around the room’s walls.
The technology was developed to ignore immobile objects and focus on signals reflected off of moving ones, reassembling a person’s posture in a radar-like image. This method worked even when a wall was placed between the routers and the people taking part.
This enabled the WiFi routers to penetrate otherwise opaque materials like cinderblock, wood, and even concrete. The latest study from Carnegie Mellon isn’t the first to develop a tracking system that can see through walls, but it does provide photos with significantly higher resolution than previous attempts. Even people’s actions can be deduced by the way they move.
To read our article about “Researchers use paper byproducts to save 100s lives with med” click here.
