Boston uses AI to reduce stop-and-go traffic by 50 percent

Boston is working with Google to optimize traffic light patterns to improve traffic flow in congested areas.

The company’s Project Green Light uses artificial intelligence (AI) to model traffic patterns and make recommendations for signal timing that can reduce stop-and-go traffic and emissions. In 2023, Boston ranked 8th in the world for the highest traffic delays.

The city has so far implemented four of Google’s recommendations at intersections in three locations by changing traffic light patterns.

“We are pleasantly surprised by the reduction in stop-and-go traffic at the pilot intersections,” said Michael Lawrence Evans, director of new technologies for the City of Boston. Cities today“A platform like Green Light is a straightforward way for us to try out more frequent changes to traffic light timing that respond to current traffic conditions. The automated impact analysis helped us to evaluate the recommendations – and thus the value of the platform – very easily.”

Over the past five months, traffic at hundreds of traffic light intersections has been analyzed using AI and Google Maps driving trends to provide recommendations for optimization.

Boston Transportation Department traffic engineers evaluated each recommendation for safety, feasibility and effectiveness to determine if the recommendation could be implemented. After implementation, Project Green Light measures the impact on traffic patterns and provides that analysis to the city.

At one intersection, stop-and-go traffic was reduced by over 50 percent, and in cities around the world using Project Greenlight, emissions were reported to have been reduced by an average of 10 percent.

“We’ve also already received some feedback from residents who have noticed a difference,” said Lawrence Evans. “In most cases, these are minor changes to the traffic lights, so it’s good to hear that people have noticed and are now having a better driving experience at these intersections.”

The city views Project Green Light, along with other AI pilot projects, as a support tool that provides traffic engineers with additional capacity to locate signals that need to be optimized, particularly at the 300 traffic light intersections that are not directly connected to the city’s traffic management center.

Lawrence Evans noted that he ultimately wants to compare the performance of AI-based signal timing recommendations with new signal schedules created in the traditional way.

Boston will continue to consider the platform’s recommendations at least until the end of this year.

Image: Gynane | Dreamstime.com