MIT helps automated vehicles see around corners

Advertisement

MIT researchers have developed a system that can sense tiny changes in shadows on the ground to help autonomous systems determine if a moving object is coming around a corner.

This work, funded by the Toyota Research Institute, could one day help keep self-driving cars from hitting other cars or pedestrians or help guide robots through hospital hallways as they deliver medicine or supplies.

MIT researchers conducted experiments with an autonomous vehicle driving around a parking garage and a self-driving wheelchair moving through hallways. in a paper presented at the International Conference on Intelligent Robots and Systems, they said the new system bested traditional lidar systems by more than half a second — not a lot, but enough to forestall accidents.

The system, dubbed ShadowCam, uses sequences of video frames from a camera aimed at a particular area, such as the floor in front of a corner. Changes in light intensity can indicate whether an object is stationary or moving.

An earlier relied simplified QR codes placed on a floor; to be used in dynamic environments, the latest version combines images registration — which overlaps images to compare for changes — and visual odometry, which estimates the motion of a camera in real time and which has been used on Mars Rovers.

As ShadowCam moves, it uses the registration method to overlay images from the same viewpoint as the robot, zeroing in on the same patch of pixels where a shadow is located, seeking any subtle changes between images.

So far, the system has only been tested in indoor settings, where speeds are lower and lighting conditions are more consistent. Researchers plan to develop the system to work outdoors as well, and to speed up the ShadowCam's shadow detection and automated the process of selecting areas to search for shadows.

“For applications where robots are moving around environments with other moving objects or people, our method can give the robot an early warning that somebody is coming around the corner, so the vehicle can slow down, adapt its path, and prepare in advance to avoid a collision,” says paper coauthor Daniela Rus, director of the Computer Science and Artificial Intelligence Laboratory. “The big dream is to provide ‘X-ray vision’ of sorts to vehicles moving fast on the streets.”

Next, the researchers are developing the system further to work in different indoor and outdoor lighting conditions. In the future, there could also be ways to speed up the system’s shadow detection and automate the process of annotating targeted areas for shadow sensing.

MIT tested ShadowCam in a parking garage. Photo: MIT