This new technology can see your body through walls

MIT's Computer Science and Artificial Intelligence Laboratory has created a system that can see your body through walls, recreating your posture while walking, sitting, or standing still.

This technology uses radio waves to sense your position, then recreates your body image in the form of simple human dolls called RF-Pose.

Specifically, researchers will utilize a neural network to analyze radio signals that bounce back from the human body, thereby creating a moving human figure capable of walking, standing, sitting and moving. move the joints when the person behind the wall performs those actions.

The team said the system could be used to monitor diseases such as Parkinson's and multiple sclerosis (MS), helping doctors better understand the progress of the disease and allowing them to adjust treatment. be suitable. It can also help older people live more independently, provide safer solutions to monitor falls, injuries and changes in daily activities.

This technology uses radio waves to sense your position.

The research team is excited to use this system in the field of health care, allowing passive monitoring of an in-room subject without the need for a camera or other potentially invasive device.

"All the data collected by the team is subject to the approval of the subject and is encrypted, anonymized to protect user privacy" - researchers said - "For real applications In the future, the group plans to apply a "consensus mechanism" in which the device installer who wants to start monitoring his surroundings will have to make certain movements to activate system".

The researchers trained the neural network by showing it a video of a person walking right next to the radio interference image they created when traveling. After that, stick human dolls will be inserted on this moving image, and the neural network will be trained to do the same thing automatically. Because radio frequency signals are so popular, its use is much easier than other sensory technologies.

More interestingly, researchers have never trained this system to look through the wall, but it was able to "generalize its knowledge to handle wall movement".

"If you think about the computer vision system as a teacher, this is a really interesting example of how students surpass teachers," said researcher Antonio Torralba. There is no information available to indicate whether this system will be widely used for other commercial purposes.