Engineers are working on all sorts of technologies that could make cars smarter and help drivers stay safe–lasers, radar, ad-hoc WiFi, and advanced image recognition are all on the table. However, these are all complicated and expensive to implement. Intel research scientist Richard Roberts has a different idea. It may be possible to use visible light emitted by LEDs to create a low-cost automobile mesh network. Many cars already have LED headlights that could be used in such a system, so we’re already off to a good start.
Intel has been working on so-called visible light communication (VLC) since 2008, but it has been on the back burner as of late. Roberts hopes that its potential as a car-to-car communication platform could renew interest. The system described by Roberts and his colleagues would use a series of rapid pulses of visible light to relay information from one car to another. These LED flashes would be so short as to be invisible to the human eye, but could potentially tell other cars about traffic conditions down the road, positioning, possible collisions, and (if you put it all together) even as part of an autonomous driving system.
The necessary data rate for sending this information from one car to another is low enough that a regular camera could be used to receive it, and LEDs already used in cars could easily relay it. However, special equipment can make VLC much more robust. That makes VLC a much cheaper form of V2V to implement than sticking radar in every new car. Make no mistake, radar and lasers are better overall technologies for making cars aware of their surroundings, but the costs are much higher, and the maximum benefits won’t be realized until at least 10% of cars on the road are equipped with such a system. Researchers estimate that’s the lowest proportion necessary to form a basic mesh network.
VLC comes with some drawback because unlike radar or wireless networking, it relies on line-of-sight. The systems being worked on right now won’t work reliably enough to be released into the wild. In bright daylight, the glare from the sun will blot out the subtle flashes from LEDs. Inclement weather like fog or snow would also make the system less effective. While this is certainly an impediment to making VLC work, it can also be a benefit in some situations. If a large number of cars have radar, for example, those signals travel outward in all directions, which can result in a lot of signal noise. The visible light flashes only relay information to nearby cars.
This might not make sense as the heart of autonomous driving technology, but VLC may be valuable as an additional layer of protection for self-driving cars in the future. Roberts is working on a set of IEEE specs that could guide future development of VLC in cars. In teh short term, it could find use in regular cars to make us safer until the robots take over the roads.