Indian origin researcher”s smart headlights that see through rainPublished On: Fri, Jul 6th, 2012 | Computer Science | By BioNews
While driving in the rain, or when there is snow fall, visibility sometimes gets worse when the headlights are on.
How would be if headlight beams could bend around the raindrops so that the driver can see what’s ahead of them.
Now, a team led by an Indian origin researcher has found a way to do just that.
By not illuminating the raindrops the headlights would avoid a common problem: in heavy rain, headlights make it harder to see, not easier. Headlight beams reflect off the rain (or snow, or fog). The reflected light heads back to the driver’s eyes, not to the obstacles on the road.
The protoype system, developed by the Carnegie Mellon University team led by Srinivasa G. Narasimhan, consists of a camera, projector, and beam splitter, linked to a computer processor, Discovery News reported.
The camera takes a picture of the raindrops at the top of the field of view. The processor can tell where the drops are headed and sends a signal to the headlights, which adjust the beams of light they send out so that there isn’t any light where the raindrop is.
The whole system is fast – the time from capture to when the light adjusts is about 13 milliseconds. A raindrop moves falls at anywhere from nine to 13 meters per second, .so the drops will only fall about 9-13 millimeters, or about a third to half an inch.
One question the team asked is how fast the system has to be –- 13 milliseconds is a bit too long given how fast most cars are moving. Increasing the range would also be good, as it currently is about 13 feet.
Making the system faster, though, shouldn’t be too big of a problem.
The protoype was built with off-the-shelf parts. Custom built equipment is usually much better integrated. Computer simulations show the system could boost the range of headlights to 90 feet.
The work was presented by Narasimhan at Microsoft Research and at Research@Intel 2012, and the results were published in the Proceedings of the IEEE Conference on Computational Photography.