Researchers at Carnegie Mellon are working on a prototype "smart headlight" that avoids shining light on raindrops and snowflakes to reduce distraction for the driver. It works using a Viewsonic projector, a beamsplitter, a Point Grey Flea3 camera, and a PC to observe and then predict where raindrops will fall. The headlight then adjusts the light output to avoid raindrops, thus making it easier for the driver to see in such conditions. The entire process from start to finish takes about 13 milliseconds.

The prototype appears to be very effective at minimizing the distracting reflections from rain drops, but researchers don't yet address snowflakes which fall differently and would be more difficult to accurately predict. Also, they do not say how well the system functions under a high volume of rain, nor how it reacts to other moving obstacles like animals, cars, or other falling objects. The technology, while impressive, still has a long way to go, so there's no word on how soon this will make its way into new cars.