“Smart” headlights improve visibility by illuminating the road ahead, but not raindrops or snowflakes.
Smarter headlights. I’m Bob Hirshon and this is Science Update.
If you drive at night through rain or snow, it’s hard to see. That’s partly because a car’s headlights light up the raindrops and snowflakes instead of the open road. Srinivasa Narasimhan and his colleagues at Carnegie Mellon University’s Robotics Institute are developing a headlight that won’t do that.
So what we are trying to do here is we are subdividing that beam of light that is going out into tiny little beams, or pixels.
In other words, it works more like a movie projector than a light bulb. A high-speed camera constantly takes pictures of the rain or snow ahead, while a computer calculates where those droplets will be in a few milliseconds. That way, the headlight can keep adjusting itself, and shine its tiny beams only into the clear areas. Right now, each adjustment takes about 13 milliseconds; Narasimhan’s team wants to get it down to two or three. I’m Bob Hirshon for AAAS, the Science Society.
Making Sense of the Research
You don't need a driver's license to know that it's hard to see through a windshield during a nighttime rainstorm or blizzard. Even from the passenger seat, you can see how poor the visibility is. As you heard, one reason is that the car's headlights, which light the way well under clear conditions, light up the raindrops and snowflakes more brightly than the road ahead.
You can see a similar effect if you've ever tried to take a flash photo of something distant in the dark—for example, of a stage performance in a darkened theater. If the flash hits something that's closer, like the back of somebody's head, it will light up the nearby object so brightly that everything else will look very dark by comparison.
This “smart” headlight works around the problem by shining light only into the clear areas, not on the raindrops or snowflakes in the foreground. It's a simple concept with a devilishly complicated solution. First, you've got to break up the light into thousands of tiny, individual beams. That in itself isn't so hard, since that's what movie projectors do—they project thousands of individual colored lights which come together to form a coherent picture.
Still, turning off all the pixels (light beams) that hit snowflakes or raindrops, while leaving all the other ones on, isn't the easiest thing to do. Then consider that the arrangement of those snowflakes and raindrops is constantly changing, as the snow or rain falls and the car moves forward. So Narasimhan had to develop a complex algorithm, or mathematical procedure, that makes it possible to use the current pattern of raindrops to estimate where those drops will be a few milliseconds later.
In order for that to work, you need an extremely high-speed camera to constantly take pictures. As soon as the picture's taken, the algorithm must predict where the raindrops are going, and then adjust the light pixels to avoid them. This has to happen, ideally, several hundred times per second.
One reason it all has to happen so fast is that the longer the time between the photo and the light adjustment, the harder it is to predict where the snowflakes and raindrops will be. As Narasimhan explains, at speeds of just a few milliseconds (thousandths of a second), the math is relatively straightforward. It's more or less a matter of understanding the effects of gravity; factors like the speed of the wind and the speed of the car don't have much of an impact. But expand that delay to just hundredths instead of thousands of a second, and those other factors begin to matter. And those elements are even harder to measure and predict, especially when you consider that they affect one another.
That's why even thirteen milliseconds isn't quite good enough to put the headlights in a car. However, Narasimhan says the smart headlights should ultimately be very safe to use, because they'll be designed to revert to a normal headlight if the computer or the camera fails.
Now try and answer these questions:
- Why do rain and snow create more visibility problems at night than during the day?
- How is the light from the smart headlights different from a normal headlight lamp?
- Why is it so important to make the computer processing speed so fast?
You may want to check out these related resources:
The Illuminations four-lesson plan Shedding the Light explores the development of a mathematical model for the decay of light passing through water.