How Smart Cities Can Beat Their Traffic Troubles

Aerial view of Abu Dhabi Corniche.

Engineers are creating a virtual replica of Abu Dhabi's road network to address urban traffic problems

Everybody talks about traffic congestion, and yet it just seems to keep getting worse. But now three NYU Abu Dhabi engineers are laying the groundwork for a new form of data-driven cooperation with police and municipal authorities, designed to move vehicles around the city more efficiently.

Anyone who’s ever lost time weaving through Abu Dhabi’s Al Zahiyah neighborhood – some still call it the Tourist Club – will wish this project speedy success.

Saif Jabari, assistant professor of civil and urban engineering, leads the “Smart Transportation” group. With Tembine Hamidou and Nick Freris, assistant professors of electrical and computer engineering, he’s developing sophisticated controls to optimize vehicular travel.

“Data is becoming more widely available,” Jabari noted, and that creates new possibilities. “We’re making a push towards real-time: when you ask your phone to suggest a route, it will take into account things like traffic congestion, and whether or not there’s an incident along the way”.

That will require comprehensive real-time data, and so the team is working toward a memorandum of understanding with Abu Dhabi’s Department of Municipal Affairs and Transport (DMAT).

We’re making a push toward real-time: when you ask your phone to suggest a route, it will take into account things like anticipated traffic congestion, and whether or not there might be an incident along the way.

Saif Jabari, assistant professor of civil and urban engineering

Today, Jabari noted, most of the world’s traffic lights in congested urban settings operate on old-fashioned fixed timing. A more advanced approach is actuated control, in which sensors, typically under the roadbed near an intersection, measure vehicle flows and regulate the light. Actuated signals are typically found along high speed arterials.

The most sophisticated control techniques are referred to as adaptive control (AC). This, Jabari explained, “looks at conditions at multiple intersections … to control traffic in a way that’s beneficial to all.”

At present AC exists as “proprietary systems you can buy off the shelf,” Jabari said. “They’re a simple algorithm, and don’t work very well,” especially when something unforeseen snarls traffic. “We haven’t reverse-engineered any of them yet,” he said cheerfully, “but we know we can do better!”

Abu Dhabi traffic models

The project’s next step is “a simulation test-bed for downtown Abu Dhabi,” Jabari said, “a virtual replica of the urban road network. We calibrate it to different demand scenarios, accidents and things like that. We plan to purchase AC devices from a manufacturer, and test other algorithms against them.”

They’re developing a fine-grained simulation down to the level of individual vehicles. Already, Jabari said, “the research is mathematically mature but we haven’t tested it in a micro-simulator … or compared it with a real-world adaptive controller.”

Those are the next steps. “I have two people on it now,” Jabari said, “and we just submitted a proposal for more manpower. We hope to complete the simulation work within a year.” After that the researchers aspire to agree with DMAT on addressing actual traffic problems with the new software. All this will lead to published papers, the engineers foresee, and ultimately to patents.