Boston partners with Google to ease traffic congestion

The city of Boston in June announced an expansion of its partnership with Google Research to use artificial intelligence as a tool to improve traffic flow and reduce congestion.

Google’s Project Green Light uses Google Maps data and AI to model traffic flow and make timing recommendations for the city’s 900 intersections. Since the city’s partnership with Google launched in 2024, Project Green Light has analyzed hundreds of Boston’s signaled intersections to identify traffic flow issues and recommend potential improvements.

These recommendations allow staff at the city’s Traffic Management Center to synchronize traffic lights between intersections to create better traffic flow and adjust light patterns to align with demand. The city has made improvements at 114 intersections using Project Green Light recommendations.

Boston partnered with traffic analytics company INRIX to assess the implemented recommendations. This work revealed a 13.5% average reduction in delays, an average 20% reduction in unnecessary stops at newly retimed intersections, and up to a 33% reduction in unnecessary traffic stoppages at congested intersections.

“Typically, signals are retimed every three to five years,” said Michael Lawrence Evans, Boston’s director of emerging technologies, “and what we found is that traffic patterns are changing more often than that.”

While city staff were already aware of traffic pattern issues during rush hour periods, they saw an opportunity to improve traffic patterns during less-busy times.

Citywide congestion led to Boston being ranked eighth in the world by INRIX for commuter hours lost to traffic delays in 2023. City staff identified Project Green Light as a potential partner due to ease of implementation and alignment with Boston’s aim to deliver more frequent adjustments through modeling.

“We didn’t have to install any new equipment,” Evans said. “All we needed to do was provide them with the location of our signalized intersections.

“They’re providing this for no cost … and it’s not something that displaces our traffic engineers. It actually works in concert with them.”

Project Green Light analyzes data from existing traffic monitoring hardware like cameras to perform its analysis and make recommendations. Google makes a recommendation once it has gathered four to six weeks of data at an intersection and follows up with an impact report a month after a change is implemented.

Staff can adjust 600 of the city’s signalized intersections remotely from the Traffic Management Center, located in City Hall, which leaves one-third of the city’s signaled intersections without the capability for real-time updates. In those cases, the Project Green Light analysis helps the city shorten the adjustment gaps, reducing the likelihood of calls from residents about timing issues.

Evans said the city has tested Project Green Light recommendations in nearly every Boston neighborhood and has seen consistent congestion reduction during the worst traffic hours. City staff review new recommendations every Thursday, which enhances the city’s understanding of traffic hotspots and the state of its equipment inventory.

“There might be a communication issue between two adjacent lights; there might be a communication issue between an intersection and the Traffic Management Center,” Evans said. “Having a better understanding of that helps us maintain the system a little bit better.”

Evans’ approach to new technology emphasizes continuous efforts to build buy-in and confidence in the project outcome.

“You can’t just sort of say here’s a fuzzy thing, here’s this AI thing that will kind of do this magic,” he said. “People internally need more detail. Once people start seeing good results, then it definitely helps them get more comfortable with it.”