(NEWSER) – If you've ever called a traffic light "dumb," well, you're technically correct. Amidst the proliferation of smart devices, traffic lights have largely remained conventional, their colors changing based on preprogrammed timing that can go years without being adjusted. But there's a noteworthy exception, and it's located in Pittsburgh. There, Carnegie Mellon University professor of robotics Stephen Smith's startup is being put to the test, and the results garnered since its 2012 implementation are encouraging. The Surtrac system uses artificial intelligence (AI) to coordinate with other traffic lights via radar sensors and cameras at each light, reports IEEE Spectrum. It's estimated to have cut emissions by a fifth and idling time by more than 40%; driving time is down 25%.
Unlike other smart systems being tested elsewhere, Surtrac's is decentralized, and its website explains the approach like so: "Each intersection allocates its green time independently based on actual incoming vehicle flows, and then projected outflows are communicated to neighboring intersections to increase their visibility of future incoming traffic." Popular Mechanics reports the lights were first installed at nine intersections; they're now at 50, and will be going city-wide. There's not yet a timeline for exporting these AI traffic lights to other cities. But as cars themselves get smarter, they'll likely be talking to the traffic lights as well, and could be alerted to, say, a soon-to-change light. Car and Driver notes the system could also give priority status to things like emergency vehicles. (These traffic lights in Germany help handle smartphone zombies.)
This story originally appeared on Newser.