By on October 21, 2016

traffic gridlock

With the exceptions of a horrible wreck or having a child in the car that you aren’t particularly fond of, nothing drains the enjoyment out of a drive like being stuck in rush-hour traffic. Every second of idling, waiting, and creeping along city streets is another agonizing moment where you could be enjoying a backroad or at home eating dinner.

Thankfully, IEEE Spectrum reports that a team of researchers is working to solve this problem with traffic lights that know all and see all.

Researchers from Carnegie Mellon University’s Robotics Institute have a startup called Surtrac that has installed traffic lights with artificial intelligence all over Pittsburgh. Surtrac — an acronym for Scalable Urban Traffic Control — collects traffic data from cameras and radar signals, allowing the network of lights to coordinate with each other to ensure flow through intersections happens as quickly as possible.

The project started with nine intersections in 2012 and has grown to fifty, unbeknownst to most commuters. The startup eventually plans to implement its network across the entire city and ultimately bring the technology to other metropolitan hubs in need of silky smooth transit.

Stating the case for the project, Carnegie Mellon University professor of robotics Stephen Smith told an audience at the White House Frontiers Conference that congestion costs the U.S. economy $121 billion every year and produces about 25 billion kilograms of excess carbon dioxide emissions.

That’s a heap of money and pollution, although it’s probably the reduction in travel time that will get most people excited. In Pittsburgh, the AI light network reportedly reduced travel time by 25 percent and idling time by over 40 percent. That translates to more time at work, more time at home, and less time cursing your own existence while you are stuck in traffic.

While other cities have complex traffic management systems, even some that adapt to changing circumstances and vehicle frequency, Surtrac is the only one where each light is responsible for its own intersection. This decentralized technique reduces the overall load on the network and makes it easier to scale up. The team has also begun working on a system to communicate directly with cars, allowing the network to notify drivers of traffic conditions in advance. It could also be used to prioritize certain types of traffic, such as emergency vehicles and city buses.

There is no set date for when AI traffic lights will be exported to other U.S. cities. But they should be a welcome addition for anyone who has ever spent the better part of their commute staring at a John Kerry ’04 bumper sticker plastered on the back of a slow-moving Prius.

[Image: Joisey Showaa/Flickr (CC BY-SA 2.0)]

Get the latest TTAC e-Newsletter!

Recommended

9 Comments on “Traffic Lights with Artificial Intelligence Could Help Your Terrible Daily Commute...”


  • avatar
    sco

    It’s a good start, but if the boys at Carnegie-Mellon could figure out how to stop people from coming to a near stop to gawk at any and every smashed car anywhere near the roadway, now that would be truly revolutionary.

    • 0 avatar
      Tandoor

      St. Louis congestion reduction plan.
      Step 1: mid highway barriers – no view of the other side of the highway.
      Step 2: if stopped by law enforcement you must exit the highway. Extra fine for stopping on the highway. Execution for stopping on the left shoulder in the middle of the highway.

  • avatar
    yamahog

    I wish the NSA would start hooking it up with all the data.

    Like giving us better speed limits, timing lights better, recommending friends, giving regular people a quick pass through the airport security, and giving lonely people a free pat down

  • avatar
    Fred

    I solved my commuting problems by living and working in a rural area. 2 stop lights, that cost me about 2 minutes. Maybe 5 if a train is crossing.

  • avatar
    stuki

    ” It could also be used to prioritize certain types of traffic, such as emergency vehicles and city buses.”

    ….And government officials. And party members. And union bosses. And those who own $150,000 cars powered by blue blood, since that is “technology” “we” have to support. And Donald Trump. In fact, by some weird “coincidence,” exactly the same people who gets invited to fundraisers for Dear Leader. And get asked for their input on the latest harebrained policy, program, stimulus, bailout…. And who gets to carry a gun around New York without fear of being arrested. How strange…..

  • avatar
    brn

    Making decisions based on input is not “artificial intelligence”. It’s an algorithm.

    Now that that’s off my chest, it’s about time! Let’s start with a couple of basics.
    1. Detect vehicles. If there are no vehicles in a particular direction, don’t rotate the green to that direction.
    2. If one particular direction is backed up, extend the green.
    3. Determine traffic flow from other lights and prepare for them by shifting the next light up or down.

  • avatar
    2manycars

    Carbon dioxide is not a pollutant and “excess” emission of same is not an issue. However, reducing traffic congestion is certainly a positive thing.

    No doubt the cameras will also record license plate numbers and perhaps run facial recognition profiles, for our protection of course. Doubleplus good!

  • avatar
    Tandoor

    In a lot of areas. St. Louis doesn’t even have basic sensors yet. After 2-3 minutes at a red with no cross traffic, the light finally changes and the green light 45 yards down the road turns red for no cross traffic. Many drivers just routinely run them.

  • avatar
    Lorenzo

    I’ve noticed that signalized intersections move traffic better when they’re flashing red in all directions than when they’re working. At least the lines aren’t as long, and people pay attention.

    America’s first traffic safety “expert” William Phelps Eno detested signals and promoted the traffic circle, aka rotary, called roundabouts in the UK. He would probably have a heart attack if he saw how the Italians negotiate them today.

    Then again, he wouldn’t be driving. He never had a license and was rich enough to employ a chauffeur.

Read all comments

Back to TopLeave a Reply

You must be logged in to post a comment.

Recent Comments

  • Ol Shel: Ridgeline is the truck most truck owners need, but are too insecure to own.
  • ajla: I guess that would depend on if you’re stretching your budget to get into a performance car. And what...
  • Inside Looking Out: And I loved three years I spent with my Lada. It had 80 hp 1.5L OHC engine. It was fabulous, the...
  • V16: It looks one hundred times better than the Hyundai Palisade.
  • Inside Looking Out: I am surprised that drivers did not try to get away from police or at least attack police when...

New Car Research

Get a Free Dealer Quote

Who We Are

  • Matthew Guy
  • Timothy Cain
  • Adam Tonge
  • Bozi Tatarevic
  • Chris Tonn
  • Corey Lewis
  • Mark Baruth
  • Ronnie Schreiber