MIT researchers develop shadow sensing technology for autonomous cars

MIT researchers have created a system called ‘ShadowCam’ which detects shadows to help autonomous cars avoid possible accidents.

MIT’s shadow sensing technology for autonomous vehicles. (MIT)

Autonomous vehicles are still a near-perfect technology. Possibly the biggest challenge for autonomous vehicles is tackling sudden or unexpected human encounters. The same could be posed by possible car collisions for autonomous vehicles. As this technology develops, MIT researchers have come up with a new system called 'ShadowCam' to help autonomous vehicles.

MIT's new safety system for autonomous vehicles aims to help avoid possible accidents with pedestrians or cars by quickly detecting when these objects are around. How ShadowCam works is by identifying tiny changes in shadows on the ground to determine if there's a car or pedestrian approaching the autonomous vehicle.

"For input, ShadowCam uses sequences of video frames from a camera targeting a specific area, such as the floor in front of a corner. It detects changes in light intensity over time, from image to image, that may indicate something moving away or coming closer. Some of those changes may be difficult to detect or invisible to the naked eye, and can be determined by various properties of the object and environment. ShadowCam computes that information and classifies each image as containing a stationary object or a dynamic, moving one. If it gets to a dynamic image, it reacts accordingly," the researchers explain.

The researchers conducted tests with ShadowCam in autonomous cars and compared it with the traditional LiDAR system. In two tests, they found ShadowCam was 0.5 seconds and 0.75 seconds faster in identifying and stopping approaching vehicles than LiDAR. Although only a few seconds faster, the researchers stress on the fact that even a second is important for autonomous vehicles.

The researchers say this ShadowCam system can be used in robots as well by giving them advance warnings to avoid hitting people that cross their path. MIT researchers aim to further develop this system and provide 'X-ray vision' of sorts to vehicles moving fast on the streets." At present, the ShadowCam is optimised only for indoor settings, parking lots in particular. ShadowCam will be developed for different indoor and outdoor lighting conditions.