AI traffic lights helped fix a Quebec town’s gridlock. An expert says Toronto may not see the same success

Professor Jeff Casello on how the Quebec town of Kirland is using AI to control their traffic signals in real time based on the flow of traffic.

A Quebec municipality is solving gridlock using an artificial intelligence-powered traffic light system that the town says is already decreasing congestion and increasing safety for commuters and pedestrians.

In February, Kirkland, a suburb on the Island of Montreal, installed the traffic lights at every intersection along Saint-Charles Boulevard – a 6-lane boulevard, the town says has seen an increased volume of traffic.

The AI-powered system -- a collaboration between Orange Traffic, a Quebec company that designs “intelligent signage” and NoTraffic, an Israeli AI company -- uses advanced AI-driven cameras, radars and sensors, to monitor traffic data at intersections.

This data gets collected and sent to computer vision algorithms for processing, which can count and detect pedestrians, motor vehicles, bicycles, and emergency vehicles, as far as 220 metres away.

The town of Kirkland reports that the technology has a “detection accuracy of 99.5 per cent.”

In an interview with CTV Your Morning, Jeff Casello, professor of transportation engineering and planning at the University of Waterloo, shared that the technology has “smarter decision-making” capabilities than traffic lights that are timed using historical traffic data.

“[The cameras] can decide how to change the traffic signal and how much green to give in each direction.”

Casello says the system, “communicates back and forth between the traffic signals throughout the network,” predicting traffic patterns up to two minutes before they happen and changing traffic lights accordingly.

According to Orange Traffic, the technology has decreased travel time along Saint-Charles Boulevard during rush hour from 20 minutes to five minutes.

Orange Traffic says those improvements will benefit the environment by decreasing greenhouse gas emissions by 2,000 tonnes per year, which is the equivalent of removing 500 cars off the road annually.

According to Kirkland, the municipality is the first in Quebec and the second in Canada to implement the AI system.

Kirkland Mayor Michel Gibson says the town “had to adapt” to increased traffic along the boulevard, which was hitting daily traffic volumes of 42,000 cars per day, despite being designed to handle only 30,000.

Without the budget to expand the road, the town turned to the AI traffic system which cost $1.8M.

Now, after seeing results in Kirkland, nearby communities, including Dorval and Pointe-Claire, are considering installing the technology. Quebec City has also begun using Google’s Green Light to synchronize its traffic lights.

But some traffic experts are questioning why larger cities are not considering implementing the system.

Casello says the AI technology is “scalable to major cities,” but doubts it could see comparable success in Toronto.

“I think we’d see benefits from this kind of system in a place like Toronto, but I don’t think you’ll see the same kind of really large increases in performance,” Casello says.

“The [traffic] volume is so high, there’s just not a lot we can do with the traffic signals to make them work better.”

The NoTraffic system has previously ran a pilot around the University of British Columbia. Now the technology is being tested for use in major cities.