Researchers at MIT have developed an artificial intelligence system (AI) that can be seen through walls.
The creators of AI, a group of scientists from MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL from Computer Science and Vision Intelligence Laboratory), report that the RF-Pose project could be used for many functions in our daily lives.
In a document substantiating the investigation (PDF), MIT reports that AI uses neural networking to teach wireless devices to sense people's posture and movement "even on the other side of a wall."
This type of X-ray vision may seem excessive, but the team, headed by Professor Dina Katabi, states that the nervous network is able to analyze radio signals bounced by people in their movement to create a digital image position and movement of the individual.
In order to prove the potential of AI, MIT presented the RF-Pose project to guess when a figure is ready to sit, stand and move its edges.
The majority of today's neural networks are trained with data manually fed into the computer. However, radio-based training was more than just a challenge.
To teach AI how to interpret these labels, scientists collected thousands of samples of wireless devices and cameras that were in places with people who had day-to-day activities. The images from the cameras were converted into digital motion and mapped to the corresponding radio signals.
The RF-Pose project was then able to estimate pose and movement without assistance cameras. But the surprise came when the researchers noticed that the new technology had the ability to see behind walls.
According to MIT, the invention may have many practical uses. For example, AI could be used for discreet monitoring of elderly people, allowing them to live independently while ensuring safety in the event of a fall or an accident.
It can also be used in medicine in the study and monitoring of the progression of diseases such as Parkinson's disease, multiple sclerosis and muscular dystrophy.
The project team is currently trying to convert the 2D figures into XNUMXD representations and is working with medical professionals to explore the applications of this technology in health care.
The research will be presented later this month at the Conference on Computer Vision and Pattern Recognition to be held in Salt Lake City, Utah.
________________________________
- Researchers built cheap touch paper
- How to report other users to the new Gmail
- Facebook: best to apologize for permission
- Konrads Voits: In prison, a hacker who has broken the prison
Slowly they use it only to take care of the elderly.
Only in a coop will we be able to have privacy and again it is not certain
I like that they made it to watch the elderly. In the end, only in a coop will we escape the monitoring and again not with so much confidence.