Weird Mayoral-Campaign Idea Evaluator: artificially intelligent traffic lights

Weird Mayoral-Campaign Idea Evaluator: artificially intelligent traffic lights

(Image: Andrzej Wrotek)

As the 2014 mayoral campaign continues, the candidates are going to advance plenty of policy ideas. Some of those ideas are bound to be really weird, whether because they’re impractical, crazily expensive, or just new and unfamiliar. In this occasional feature, we’ll pick a few of those types of proposals and weigh the odds of them ever actually happening.

What It Is: If it worked as advertised, this idea would be a magic bullet—a way for some lucky mayor to dramatically reduce gridlock virtually overnight (and, naturally, take all the credit for doing so).

There are other technologies that can supposedly make traffic lights smarter, but the one that keeps getting mentioned by Toronto mayoral candidates is called MARLIN-ATSC. The reason this particular system has become such a hot topic locally is that it’s being developed at the University of Toronto, by a team of researchers led by professer Baher Abdulhai and an engineering post-doc named Samah El-Tantawy.

The details are very technical, but the layperson’s explanation goes like this: MARLIN-ATSC uses sensors and computer processors to link traffic lights at different intersections, allowing them to “think” as one. Rather than operating on timers or reacting to pre-programmed instructions, MARLIN-enabled lights adjust the length of reds and greens in response to real-time data about traffic flows. The system can even make itself smarter, by fine-tuning itself automatically over time. In theory, the amount of human intervention needed to optimize Tornoto’s intersections would be minimal. The researchers claim their system can reduce intersection delays by 40 per cent.

Who’s Proposing It: Karen Stintz made the biggest splash with her proposal, but the system has also been name-dropped by David Soknacki, and John Tory met with researchers for a demonstration. Olivia Chow has promised to speed the implementation of “smart traffic lights,” but hasn’t mentioned MARLIN by name.

The Prognosis: There are a few things working against MARLIN. For one, city staff don’t seem sold on the idea. Steve Buckley, Toronto’s general manager of transportation, told the Globe that he’s skeptical of the 40-per-cent delay-reduction claim. Even Abdulhai, the U of T professor who’s helping develop the system, admits that it has only been tested in simulations and may not perform as expected in the field. “When you put things in the field, there is always potential for surprises,” he told us. “We want to be surprised.”

The city’s traffic managers aren’t sure they want Toronto’s streets to be a testbed, though. Transportation staff are talking to MARLIN’s developers, but they’re also looking at similar types of systems from other providers. About 350 intersections already have some form of computerized real-time signal adjustment, but the system is 20 years old, and it’s showing its age.

On the plus side, Stintz has said that she has already secured $200,000 towards a 10-intersection MARLIN pilot project (half from Metrolinx, and half from U of T). If city council could be convinced to throw in a paltry $100,000 of its own, MARLIN could start collecting valuable real-world data, which would go a long way towards determining whether or not it’s the real deal.

Odds of it Ever Happening: 5:1. There’s lots of enthusiasm for using homegrown tech to tackle gridlock, and $300,000 isn’t much of a barrier to entry. There’s no way of knowing if the system would deliver the promised results, though.