Drive on a city’s streets at night and you’re guided by artificial lights: glowing traffic signals beckoning you forward, the headlights of a car trailing you, a sign warning of work ahead.
Artificial lights may soon guide your car, too: In the quest for cars that understand the world around them and respond intelligently, a growing number of research engineers are exploring systems that encode signals in LED light.
“We envision car lights transmitting messages that your eyes can’t see,” says Richard Roberts, a research scientist at Intel. “They’re blinking out messages to be used by other automobiles for safety reasons: positioning, collision avoidance, cooperative driving—maybe even someday for autonomous driving.”
Roberts has been a part of this work since Intel started looking into visible light communication (VLC) in 2008, and he’s seen it go from the “next Wi-Fi” to a research topic on hold—at least at Intel. While companies and researchers have tried using visible light to send extra information from a sign when scanned, as with a QR code, to pinpoint locations within a store, and even as a high-bandwidth Wi-Fi substitute—called Li-Fi—it just hasn’t caught on in the mainstream.
Now Roberts and other researchers think they’ve found its killer app: car-to-car communications.
VLC is based on the idea that while humans can’t pick up on fast modulations in light, cameras can. Unlike Li-Fi, which uses special receivers, versions of VLC for cars rely on data rates low enough for ordinary cameras and would use the car’s existing lights, making such a system inherently inexpensive. The lack of purpose-built equipment also distinguishes the effort from today’s radio-frequency communication standard for vehicles.
“Adding more hardware onto a device is always an issue,” says Roberts. “But cars are becoming full of cameras.” In fact, the United States will require back-up cameras to be installed in new cars beginning in 2018. And many cars already have LED headlights that can be used for transmitting. “So we said, pragmatically, what can you do with that? It turns out you can do quite a bit.”
Roberts and others are working on ways to communicate details about a vehicle’s size, speed, and actions on the road so that the cars around it can calculate its exact location and be aware of its behavior. Such systems could warn of impending collisions, help judge whether lane changes are possible, and help coordinate cooperative adaptive cruise control, where the cars signal to one another so they can respond quickly to upcoming changes in traffic flow.
Hsin-Mu Tsai, a computer scientist at National Taiwan University, has been working on a prototype VLC device to increase the safety of the cheap motor scooters that are common on the streets of Taiwan. Apart from its affordability, Tsai says that one of VLC’s greatest advantages is something that initially sounds like a problem: the fact that it works only over line of sight.
“Radio-frequency communication is usually omnidirectional,” says Tsai, “and it covers a lot more than you need.” Such signals can interfere with each other when it gets too crowded and can lead to confusion over where the signals are coming from. With VLC, the only transmissions your car picks up are the ones adjacent to it—the ones it needs to know about most for safety.
Another bonus is that a camera measures an entire scene, so it can receive several signals at once and distinguish among them. And the location information helps assign objects to signals, according to Tsai: “The beautiful thing about VLC, when we’re using a camera as a receiver, is that the actual pixels are transmitting information to you.”
Of course, cars are rarely cruising around in a perfectly clear and dark environment. In order for such a communication system to work, it has to withstand all the variations of weather and sunlight a driver could encounter.
“Due to the sunlight noise, the operation of VLC at daytime is the biggest challenge in vehicle-to-vehicle VLC development,” says Sung-Yoon Jung, an assistant professor in the electronics engineering department at Yeungnam University, in South Korea. Jung has been modeling and testing VLC protocols to identify their flexibility and performance. Some protocols allow signals to be received from as far as 30 meters, even in daylight. And a 2013 study showed that VLC receivers are more sensitive than human eyes.
VLC’s proponents say that their next mission is to get the technology incorporated into cars. Roberts and others are working now to develop a series of IEEE standards for the technology to coordinate communication efforts and help convince automakers to pick it up. If adapted widely enough, VLC could be incorporated into infrastructure such as traffic lights as well.
According to Tsai, about 10 percent of cars need to be transmitting for any vehicle-to-vehicle system to be effective, which would take around two years if every new car included the technology. But it’s a saturation nobody has begun to reach—even the dedicated short-range RF standard for automobiles is not yet in use.
Sven Beiker, the executive director of Stanford’s Center for Automotive Research, says it’s clear that many technologies, including VLC, would work in car-to-car communications. “It does not seem to be a technology question,” says Beiker. “It’s this infrastructure, this standardization, and this monetary question. It’s tough to find a business case for vehicle-to-vehicle communication.” Beiker believes that business case might be to enable self-driving cars. The two technologies are complementary, he says.
This article originally appeared in print as “Hidden Messages in Headlights.”