MIT technology sees through the walls

It works with wireless signals and tracks movements of people

      Researchers from the Laboratory of Computer Science and Artificial Intelligence (CSAIL) at MIT have created a neural network that analyzes the radio signals that bounce off people's bodies, and then they can create a dynamic gure that walks, stops, He sits and moves his limbs as the person performs those actions.

The team notes that the system, called RF-Pose, could be used to monitor diseases such as Parkinson's and multiple sclerosis (MS), providing a better understanding of the progression of the disease and allowing doctors to adjust medications accordingly.

       It could also help older people live more independently, while providing the added security of monitoring falls, injuries and changes in activity patterns.

      The team is currently working with doctors to explore multiple applications in health care. "We have seen that monitoring the walking speed of patients and the ability to perform basic activities on their own gives health care providers a window into their lives that they did not have before, which could be significant for a range of patients. diseases, "says Dina Katabi, creator of the device she sees through the walls, and has been awarded by it.

In addition to health care, the team says that RF-Pose could also be used for new ranges of video games where players move around the house, or even on search and rescue missions to help locate survivors.

One challenge that the researchers had to address is that most neural networks are trained using hand-tagged data. A neural network trained to identify cats, for example, requires people to observe a large set of image data.

To address this, the researchers collected examples using both their wireless device and a camera. They gathered thousands of images of people who perform activities such as walking, talking, sitting, opening doors and waiting for elevators.

Then they used these images from the camera to extract the gures from bars, which they showed to the neural network together with the corresponding radio signal. This combination of examples allowed the system to learn the association between the radio signal and the gures of the people in the scene.

After training, RF-Pose was able to estimate the posture and movements of a person without cameras, using only the wireless reflexes that bounce off people's bodies.