Seeing Through the Wall

The ability to look right through a wall is a distinctly uncomfortable thought and so far, the stuff of science fiction. Nevertheless, such 'X-ray' vision could have important uses, which is why researchers have been trying to get a good peep for some time. Now, MIT's Computer Sciences and Artificial Intelligence Laboratory might be on to something.

Led by Professor Dina Katabi, the project called RF-Pose is geared towards gaining a better understanding of people's posture and movement and could help the elderly who are prone to falls and injuries or benefit those suffering from conditions that affect motor coordination such as muscular dystrophy or Parkinsonism. It is thought that monitoring their movement could ensure timely assistance and help make fine adjustments to medication.

The research leverages the fact that wireless signals (radio waves) in Wi-Fi frequencies can pass through walls and reflect off human bodies. According to MIT News, the team used a neural network to analyse such radio signals and utilised artificial intelligence (AI) to teach wireless devices how to sense a person's movement from those signals.

To start with, the radio signals bouncing off a person, also known as radio frequency (RF) heat map or confidence map, can be converted into a wireframe-like figure that moves in the same way the person does. Next, researchers got thousands of images featuring people in various movements to extract wireframe figures and these, along with corresponding radio signals, were used to teach the system the correlation between RFs and people's movements. Post-training, RF-Pose was able to estimate a person's posture and movement without cameras, using only the wireless signals that bounce off the human body.

Katabi says the radio wave-based system is almost as accurate as the visible scenes. "We've seen that monitoring patients' walking speed and ability to do basic activities on their own gives healthcare providers a window into their lives that they didn't have before, which could be meaningful for a whole range of diseases," MIT News reports, quoting Katabi. Potential uses could also include search and rescue work and police surveillance.

People participating in the project have consented to data usage. However, the project data has been anonymised and is not traceable to any individual. Of course, one cannot say what privacy implications the technology would have if it is made accessible to mainstream users.