Does technology have a solution to the world’s traffic problem?

Well, flying drone-taxis and driverless cars will save the future, but in the meanwhile, is there another technology that can help us?
29 August 2018

AI can solve the traffic problem for some of the world’s most congested cities. Source: Shutterstock

Nobody likes being stuck in traffic — it’s inefficient and unproductive. However, it’s a reality of life and most of us have to deal with it, at least twice a day.

Whether you live in London, New York, Los Angeles, or Boston, your city’s roads are congested during peak hours and holidays, without fail. It’s a problem that city dwellers have created for themselves, and the pollution from fossil fuels is aggravating the situation.

People often say that new technologies such as flying drone-taxis and autonomous cars will solve the problem. Unfortunately, they’re not here yet. And when they make a debut in the market, mass adoption might take a few years.

In the meanwhile, is there a solution that can tackle the problem without requiring too large an investment, quickly and effectively? Some experts think there is.

Can artificial intelligence solve your traffic woes?

Well, artificial intelligence (AI) can be the solution to your city’s traffic problem.

Companies such as IBM, Siemens, and many others are working on AI-powered traffic lights that could monitor, evaluate, and manage traffic more appropriately.

In fact, the United States Patent and Trademark Office recently approved a patent application IBM filed in January last year for its Cognitive Traffic Signal Control System.

Of course, mechanisms and algorithms developed by each company will be different, but the basic idea is quite simple: A computer system monitors traffic video feeds in real-time and determines the impact of traffic from a certain direction. Based on its computations, it determines and directs traffic in directions that minimize congestion.

Do AI-based traffic signals actually work?

AI-powered traffic signals work. There’s evidence.

Last year, Milton Keynes in the UK and Pittsburg in the US embarked on a journey to use AI-powered traffic signals.

According to a local news story, Pittsburg’s AI-powered traffic signals helped reduce travel time by 25 percent, braking by 30 percent and idling by more than 40 percent.

The story, however, reveals that the cost to put such a system in place costs US$20,000 per intersection. Pittsburg has 600 intersections.

The total cost of installing the system will need the municipal authorities to cough up a cool US$12 million. Not a lot of money? Well, that’s because that’s not a lot of traffic intersections. New York City, on the other hand, has 12,460 intersections, and deploying the solution will cost the government US$249.2 million.

At Milton Keynes, the cost to put the system in place will be GBP3 million (US$3.86 million) according to The Telegraph, and will go live next month.

Scaling it up to other cities such as London, Edinburgh, and Manchester might be equally expensive, although there is news that the government is keen on trialing the solution in different pockets of the country.

When will the world get AI-based traffic signals?

For now, it looks like most cities will have to make do with conventional time-cycle based traffic signals. However, they could draw inspiration from Asia where Alibaba is making headway and making a mark in China and other parts of the region with its AI-powered traffic solutions.

The company deployed its solution in Hangzhou in China and reported that its solution resulted in an average traffic speed increase of 15 percent, and boosted reporting traffic violations with 92 percent accuracy. Later, it was rolled out in other parts of China — and earlier this year, was implemented in Kuala Lumpur, Malaysia.

Siemens too, deployed its solution in Banglore, one of India’s most congested cities, earlier this year, and expects to deploy its solution in other cities across the country.

The traffic situation in the US and in the UK isn’t getting any better, so maybe there’s a lesson for the rest of us to learn from the use cases in Asia. Maybe we need to move towards AI-based traffic lights sooner.