RF-Pose Sees Through Walls with AI

X-ray vision has firmly been embedded in comic books, science fiction, and joke-store devices that work as well as their price tags would indicate. Now, however, researchers at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) has taken this imaginary tech one step closer to reality with the “RF-Pose” project.

The project senses changes in radio frequencies when a human comes into view, and uses an AI trained with both images of humans in certain poses and the corresponding reflected radio frequencies from their bodies to tell what someone is doing. The image seen above illustrates this ability quite well, where stick-legs are generated on a human whose torso is visible through a window. While normal humans can’t see through walls, the AI system was trained using images and RF changes with no visible barrier, but was then able to pick out people’s poses when a wall was placed between them and the system.

Additionally, RF-Pose can be used to identify particular individuals by their RF signatures, potentially useful in a search-and-rescue operation. While applications for this technology would be wide-ranging, researchers point to medical applications where a patient could be more closely monitored than previously possible.