EnLighten App Uses AI to Predict When Lights Will Turn Green

Jobs from Indeed

Silicon Valley is looking for guinea pigs, and not to beta test a new compression algorithm. Test subjects will need a body, sure, but theyll also need an iPhone 4 or newer model and a car, preferably a BMW.

Welcome to the living lab of San Jose, where a pilot study with real-time traffic data company Connected Signals could yield a much smarter commute.

About 50 local drivers are now zooming about the Valley armed with Connected Signals EnLighten app (which BMW is building into new models). Thanks to machine learning and predictive analyses, these apps let drivers in on a little secret: When the lights will turn green.

That red light will turn green by the time you reach it, the app might tell you, by way of a speedometer interface—their signal to keep on trucking at the speed limit. Or it might tell you you’ve got no chance of making the next light, so you should start slowing down now. And when you are sitting below the red circle, you’ll know how long until it goes green.

No more green-yellow-red ternary. More data means a lot more information about how driving works. And Connected Signals is far from the only one in the traffic data game. Sector giants like Waze and Inrix make truckloads of money off collecting, organizing, and selling traffic data to governments and private businesses.

Now for the real challenge: Figuring out how to make that information useful for smartphone-equipped users. And maybe, also, for the planet at large.

With its pilot, San Jose is trying to determinetwo things. 1) Does an intelligent traffic signal app actually make driving safer? And 2) by discouraging idling and stop-and-go traffic, could it actually cut down on the communitys pollution? The Argonnne National Laboratory, operated by the University of Chicago on behalf of the Department of Energy, will look at the completed data and help the city (and Connected Signals) figure out whats up.

And should you happen to be among the elite living in this rarefied part of California, San Jose needs about 350 more guinea pigs to join the study.


To give drivers the inside scoop on traffic signals, Connected Signals must first find a city with a relatively advanced and connected traffic signal system. (San Jose updated its setup a few years ago, so its good to go.) With raw data in hand, the first plugs the behavior of each light into a learning model that takes current traffic conditions into account. It tests each model for each light until it knows it has an accurate prediction of how the light works. Then it feeds that info back into apps, through the cellular network.

Giving drivers more information, like a countdown, will help people who want to be safe, says Scott McIntyre, a human factors psychologist with Arizona State University Colleges at Lake Havasu City. But like a sword, it can be used for good or evil. People who are in a hurry and are willing to take on more risk will use that information for their goals.

On the left, the EnLighten app shows driving at the current speed of 17 mph will get a driver through the next two green lights, though the closest light is currently red. On the right, the app shows the red light up ahead will stay that way for approximately 21 seconds. Connected Signals

Connected Signals has worked with engineers and psychologists to make the app as safe as possible, says David Etherington, its co-founder and president. (Thats why the interface looks like a speedometer, he says—drivers are already very comfortable with lookingat dials on the go.) At red lights, the appceases its countdown and sounds a pay attention chime about—but not exactly —five seconds before the light turns green, to force drivers to look at the road. And the app won’t tell drivers when accelerating would let them clear a green.

When you see a green light in front of you, you tend to speed up, regardless of other factors, says Etherington. If you tell drivers theyre not going to make the light, theyre not going to speed up.

Potential San Jose test subjects shouldnt be too nervous. About 100 municipalities already give Connected Signals access to their signaling data streams, and half of those open up the firms number crunching to the public, giving individual cars access to predictive traffic light tech.

It also just might work: A study lead by Virginia Tech civil and environmental engineer Hesham Rekha, to be presented at the Transportation Research Board conference early next year, indicates sending signal countdowns and recommended speeds straight to drivers (albeit only through audio signals and not visual cues) could really, truly cut down on fuel consumption. Our (more) emissions-conscious German researcher friendsseem especially interested inusing advanced driving systems to fight pollution.

For its part, Connected Signals isn’t making money directly off its apps, or the cities that decide to use them. The company raises funds by selling data to car manufacturers and other data providers, like folkstrying to make better and more accurate maps. “We dont want cites to be in a positionwhere only someone with latermodel cars can use the app,” says Etherington.

Its an appropriate theme for Silicon Valley: Your data (anonymized) is the product. Hopefully, you’ll get a safer, smoother, pollution-free ride in exchange.

Read more: https://www.wired.com/2016/10/enlighten-app-uses-ai-predict-lights-will-turn-green/