Can Artificial Intelligence (A.I.)-infused intersections make your daily commute more tolerable?

Felix Chuang

Chances are you’ve found yourself sitting in vain at a red light with no cross traffic anywhere in sight or, evetrafficlightn worse, having to wait through three streetlight cycles (or more!) when the green light is of an impossibly short duration given the volume of traffic waiting to get through. “How could municipal planners possibly be so thickheaded?!,” you’ve probably yelled, to nobody in particular, as the light goes red again for the car in front of yours. And the wait continues.

If this happens to you on a regular basis (i.e., during your daily commute), then this post is for you. In one of the more imaginative, and universally useful, applications of the Internet-of-Things we’ve seen, a startup in Toronto is betting it can shave 40 percent off the time you wait, on average, at red-lighted intersections. Which is to say, if your commute takes you through 30 intersections en route, that means 12.5 fewer minutes in the car.

The system’s developer, University of Toronto Assistant Professor, Dr. Samah El-Tantawy, took inspiration from watching snarled traffic in Toronto and in her hometown of Cairo, Egypt. It goes way beyond existing traffic monitoring systems that use embedded sensors in the pavement to tell individual lights when to go red and green based on a car’s presence because, obviously, there are times of day when cars are omnipresent and such a simplistic approach will fail.

Instead, El-Tantawy says the system is a “brain for the traffic light” and relies on game theory to draw conclusions and make decisions.

“The [traffic lights] act as a team of players cooperating to win a game — much like players in a soccer match, where each player endeavors to score, but at the same time considers the ultimate goal of the entire team which is winning the match. Each traffic light — or agent — makes a decision every second about the best way to keep motorists waiting for as short a period as possible. The agents learn, until they converge, with each one getting the best response action to achieve its goals, without negatively affecting the others. The decisions by agents affect each other, so it’s a game.”

In practice, the proposed system is a coming together of remote data collection, data processing and peer-to-peer device communication whereby intersections gather camera-based data from the four roads converging and from actual cars on the road approaching, then share it with neighboring or adjacent intersections. The idea is to use that data and continually minimize the number of cars approaching and waiting at any given light.

Sounds fascinating for sure, but could it really work? Here’s the rub: it is already beyond theory. Tests of the system on 60 downtown Toronto intersections at rush hour showed a reduction in delays of up to 40 percent. The test also showed it cut travel times by as much as 26 percent.

The estimated cost of $20,000 to $40,000 per intersection is a reasonable price tag considering full-scale infrastructure renovations and expansions that would otherwise be required to address long-term traffic congestion. Add to that the sheer emissions savings and less “non-value” time spent for everyone, and I am of the opinion this could be one of the more noteworthy IoT examples ever to take hold.

Let’s applaud this successful test and monitor closely to see if Dr. El-Tantawy and her team can capture the attention of leadership in our most traffic-affected cities.

Subscribe Here!


see all