Artificial Intelligence Lets You See Things Hidden Around Corners

Researchers at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) have developed an artificial intelligence that can see things hidden around corners – a breakthrough that has implications for many tasks, from emergency response to self-driving cars and military applications.

The artificial intelligence developed by MIT researchers dubbed as “CornerCameras,” which works with an ordinary smartphone camera, uses data about light reflections to see people or objects that are hidden and calculate in real time their trajectory and speed.

The researchers specifically utilized the small amount of light reflected by objects on the ground referred to as “penumbra.” Utilizing the video of the penumbra, MIT’s CornerCameras create a series of one-dimensional images that tells information about the objects or people hidden around the corner.

“Even though those objects aren’t actually visible to the camera, we can look at how their movements affect the penumbra to determine where they are and where they’re going,” Katherine Bouman, lead researcher of CornerCameras, told CSAIL News. “In this way, we show that walls and other obstructions with edges can be exploited as naturally-occurring ‘cameras’ that reveal the hidden scenes beyond them.”

Bouman said that a tool that enables humans to see around obstructions would be beneficial for many tasks, from assisting firefighters to find people in burning buildings to helping self-driving cars detect pedestrians in their blind spots.

According to MIT researchers, CornerCameras still has some limitations. As this technology uses light, the absence of light in the scene and in the hidden scene itself will render it to be useless.

This technology also falters when light conditions change, for instance, when clouds are continuously moving across the sun. And due to the limitation of smartphone cameras, detecting hidden things around the corner gets weaker as one moves farther away from the corner.

The researchers said they plan to address the above-mentioned limitations in future endeavors.