The future of automotive headlamps

Programmable Automotive Headlights, Carnegie Mellon Illumination & Imaging

Nearly three years ago, a team from Carnegie Mellon’s Robotics Institute, Mines ParisTech, and Texas Instruments, approached the problem of driving in the rain, with a brilliantly simple idea.

We’ve all been there – come nighttime, your headlights shine as brightly as ever, but at the rain, not the road. The road is still dark, but your eyes are adjusted for bright light.

Their idea is to simply project light precisely around the rain.

Sound hard to you? Well, it turns out that someone’s already solved the hardest part of the problem:

Much easier than using a lampshade, or some other screen

Author: Michael Movchin
Licensed under the Creative Commons Attribution-Share Alike 3.0 Unported license.

In fact, a standard projector is just as bright as ( if not brighter than ) an average automotive headlamp.

The rest of the work for a proof-of-concept is fairly straightforward. Researchers from the University of California (Santa Cruz ), Princeton, Universitat Ulm, and MPI Informatik, provide the theoretical/physical basis for this system.

Develop software that can:

Identify the brightest regions of an image (i.e. raindrops/snow)
Predict/track the locations of the detected objects
Generate a bitmask of the raindrops
Project a bitwise-NOT of the raindrops.

For those without a background in computer graphics, the result is bit like the right side of the image below:

The result is fantastic:

But that’s just a proof of concept. Nowhere near an actual working prototype!

In the three years since then, they’ve been hard at work. With grants from the Intel
Science and Technology Center for Embedded Computing, the U.S. DOT, the Office of Naval Research, and a gift from Ford, they’ve build a working prototype.

This prototype reacts thirteen times faster, and has 143x the resolution ( 29280 pixels vs 4194304 pixels ). Additionally, they’ve added anti-glare: