New AI - Seeing through Walls
|MIT's RF-Pose "See-thru AI"|
The project, as pictured above, is called RF-Pose. The tech is breakthrough. The neural network analyzes the radio signals bouncing off the person's body as they move around and stop in enclosed rooms. The AI builds a map with an active stick figure mirroring the person's live motions.
Different AI Strokes
This AI program is very different. Usually neural networks are trained on massive sets of data. In this case, the researchers had to train the AI to recognize radio signals bouncing off people.
To do this, they used their wireless devices to collect images of people in motion and the signals bouncing off of them. From that, the researchers created stick like figures and fed that along with their signals to the neural network.
Big Surprise - AI Overachiever
The system worked and even worked on the other side of walls. That was a big surprise. The researchers had never trained the system with data collected from the other side of walls. As one MIT researchers put it, the AI student really outperformed the research teachers.
Multiple Uses for See-Through
Next step for the researchers is 3D stick figures that can detect even smaller body movements. And, the potential uses of this AI technology are many. The MIT researchers believe the health care uses could be revolutionary. For instance, it could help doctors better understand, monitor and treat progressive diseases like MS, muscular dystrophy and Parkinson's. Other applications include search and rescue and child and elderly care.