There has been a surge in active vehicle safety features in recent years, and they’re doing wonders for traffic safety. These systems, such as pedestrian avoidance, adaptive cruise control, and automatic braking and steering, protect both drivers and pedestrians by taking advantage of the virtually instant reaction times of computers. Even when you’re at your most alert, you simply can’t react to a dangerous situation as quickly as a computer that’s checking the car’s surroundings thousands of times a second.
But, as quickly as that computer may be able to react, it still needs information to react to. Some scenarios present a unique challenge, such as when you’re approaching a blind intersection and the car’s cameras can’t see cross traffic because of an obstruction (like a building). Current systems can’t respond to another vehicle crossing your path until it comes into view — at which time it could be too late to avoid an accident.
That will change soon, with a new algorithm developed at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL). The algorithm allows a computer to actually detect objects around a blind corner by measuring the reflected light on the ground. By measuring the properties of this reflected light — it’s color, intensity, and movement — the algorithm can make accurate inferences about what’s around the corner.
The potential for saving lives here is obvious: the more warning your car’s active safety systems have about an approaching object, the better the chances of it responding in time to avoid an accident. It would, for instance, be able to detect a speeding truck before you’re able to see it, or a pedestrian running on the sidewalk towards the road. And, the algorithm doesn’t require any special hardware, and could easily be integrated into existing vehicle camera and safety systems.