Google has introduced its AI system, Project Green Light, to 14 cities, aiming to optimize traffic light coordination and reduce stop-and-go traffic by 30%. Juliet Rothenberg, the product lead of Climate AI at Google Research, revealed that this initiative, now in its second year, seeks to facilitate consecutive green lights or “green waves.”
Findings from the University of Michigan indicate that such systems could also be economically advantageous for traffic management agencies. These solutions leverage existing infrastructure by using real-time data from internet-connected vehicles and navigation apps to adjust signal timings dynamically.
Vehicles as Traffic Sensors
Henry Liu, who heads a research team at the University of Michigan, explains that cars act as mobile traffic sensors by transmitting trajectory information. This method promises to cut down the waiting time at over 300,000 traffic signals nationwide, with cities like Abu Dhabi, Hamburg, Seattle, and Kolkata seeing a 30% reduction in congestion where Green Light has been implemented.
According to a 2021 study by traffic-analytics firm Inrix, Americans spend about 10% of their travel time idling at intersections during short trips. Reducing idling at traffic lights not only improves travel efficiency but also has environmental and health benefits, as pollution levels at intersections are drastically higher compared to open roads.
Data-Driven Traffic Management
Google enhances traffic-light timing using data from Google Maps users. Additionally, University of Michigan researchers used telemetry from connected General Motors vehicles to help Birmingham, Michigan, assess the timing of 34 traffic signals in 2022. These minor adjustments led to notable improvements during peak travel times, said Oakland County's traffic safety director, Danielle Deneau.
Liu notes that traditional traffic signal timing studies are costly, approximately $5,000 each, and thus are rarely updated. Many intersections have not had their timings reviewed in years, making Google's real-time data and recommendations a more financially viable alternative.
Fixed vs. Dynamic Signals
Currently, most traffic signals operate on fixed schedules for peak and off-peak hours. Dynamic, connected signals that adjust based on real-time data are costly to implement and maintain, with modernization costs per intersection exceeding $250,000, plus annual maintenance expenses of $5,000.
When cities implement Green Light, they receive a dashboard with suggested signal timing updates. Google's AI system also tracks the effects of these adjustments, enabling cities to evaluate actual outcomes. In Seattle, the system advised updates at five intersections beginning in June 2022, leading to lasting modifications at four of them.
Green Light has a waitlist, but Google aims to extend the project to additional cities later this year, Rothenberg confirms. Liu highlights that other technology startups or transit agencies could replicate this approach, given that the methodology is accessible. The primary requirement is obtaining timely data from automakers or navigation apps.
