Artificial intelligence allows to “see” through walls
Massachusetts Institute of Technology researchers have developed a system with which they can “see” through walls. A neural network called RF-Pose analyzes the radio signals of Wi-Fi devices that are now in almost every building. The signals bounce off human bodies and thus the algorithm can determine the posture of specific people.
In 2013, the teamo³ Massachusetts Institute of Technology (MIT) has found a wayob to „see” Through walls using the signaloIn radio Wi-Fi. Now the teamoł scientistow under the direction of Dina Katabi of the MIT Computer Science and Artificial Intelligence Laboratory introduced the latest version of this technology, called RF-Pose. Researchers have taught a neural network to infer a person’s attitudes and movements.
Artificial intelligence uses radio signals to use them to locate and track people even when they are completely invisible – are on the other side of the wall. The neural network analyzes radio signals, whichore bounce off human bodies, then it can create a dynamic image of a person whoora moves or gestures.
The neural network learned „vision” through walls by porownnaging thousands of scansow radios and images showing the same examples of motion or posture, including walking, morally, opening doors or waiting for an elevator.
Scientists using their smartphoneow have collected thousands of images of people doing the activities mentioned above. They then used these images to extract figures thatore showed the neural network along with the corresponding radio signal. This combination exampleow allowed the system to learn the link between radio signals and people’s figures.
After such training, the RF-Pose system was able to determine the posture and movement of the waspsob using only signal analysisoin radio signals bouncing off the body. Oprocz motion detection, the authors showed roalso that they can use wireless signals to accurately identify a specific person’s.
RF-Pose system could be useful in monitoring waspsob sick or osob in advanced age. It can provide additional security by picking up fallenow, traumaow and changes in activity patterns. – Headohe major advantage of our approach is that patients do not have to wear a sensorow nor remember to charge their devices – said Katami.
RF-Pose could also find applications in video games or even in search-and-rescue missions to helpoc locate individuals whoore survived at least an earthquake. Intelligence agencies will also enjoy the solution developed at MIT, as it is an excellent tool for tracking.