Manipulating vehicle speeds through techniques known as “eco-driving” could cut carbon emissions from idling and accelerating at intersections by 11 to 22 percent, according to a new study.
The findings highlight the surprisingly hefty climate impact of stop-and-go traffic, which may be responsible for as much as 15% of emissions from land transportation in the United States. That’s equivalent to half of U.S. airline sector emissions.
Most traffic control measures are implemented through infrastructure such as stop signs, traffic lights, speed bumps. Eco-driving, in contrast, is a set of techniques aimed at modifying driving patterns and behaviors by making changes at the level of individual vehicles.
Such measures could include apps or vehicle dashboard displays to guide speeds or eventually programming semi- or fully autonomous vehicles to automatically speed up or slow down.
“Eco-driving is an impactful, practical, and near-term intervention for climate change mitigation,” says study team member Vindula Jayawardana, a research scientist at the Toyota Research Institute who conducted research while a graduate student in lab of Cathy Wu at MIT. “It’s a reminder that small changes to individual driving behaviors, when scaled up, can make a meaningful difference in addressing transportation decarbonization.”
Eco-driving has intrigued researchers for nearly four decades, but most past studies have focused on how to implement eco-driving. This is the first large-scale study to calculate the potential climate impact of eco-driving.
It has been difficult to get a handle on that in the past because it’s such a huge computational problem: there are so many and varied traffic scenarios to model. Existing studies are small-scale, varying in assumptions and thus results.
In the new study, researchers used an AI method called deep reinforcement learning to assess the potential emissions savings from eco-driving in three US cities: San Francisco, Atlanta, and Los Angeles.
They identified 33 factors that influence vehicle emissions, ranging from temperature and road geometry to vehicle age and driver behavior. Then they created digital replicas of more than 6,000 intersections and simulated more than 1 million traffic scenarios using an open-source traffic simulator.
The work represents the first city-scale study of eco-driving’s impact, and is nearly four orders of magnitude larger than previous studies in this area.
Adopting eco-driving throughout a city could cut carbon emissions at intersections by 11 to 22 percent—without affecting traffic speed or safety, the researchers found. The results appear in the journal Transportation Research Part C: Emerging Technologies.
The benefits of eco-driving vary by city. The approach is not as effective in denser cities like San Francisco due to smaller distances between intersections. In contrast, Atlanta would see greater benefits because of higher speed limits.
Still, the benefits could really add up. Extrapolating to cities nationwide, the potential emissions savings from eco-driving across the United States are equivalent to the national emissions of Israel at the low end or Nigeria at the high end.
But full rollout isn’t necessary to see substantial benefits from eco-driving, the researchers found. If just 10 percent of vehicles on the road were eco-driving this would achieve one-quarter to one-half of the total potential emissions benefits. This is because the eco-driving vehicles would act as “pace cars,” smoothing out the traffic flow behind them.
As well as sending messages to smartphone apps or autonomous vehicles, another way of achieving eco-driving is to dynamically modulate posted speed limits at intersections. Implementing this strategy at just 20 percent of intersections would achieve 70 percent of the potential emissions cuts. This means that eco-driving measures could be rolled out gradually and still have benefits.
But here’s the rub: The specifics of how to modulate traffic depend on how many cars are eco-driving. For example, longer green lights hinder emissions reductions at low adoption rates but enhance them at high adoption rates. “The high-impact intersections shift as adoption levels change, which means careful planning is key,” Jayawardana adds.
Despite these complexities, the researchers argue that eco-driving is a good strategy for reducing emissions because it rests on a technology just as ubiquitous and quotidian as stop-and-go traffic: the smartphone. And common technologies build on each other, the researchers’ calculations reveal. “While not entirely surprising, it’s encouraging to see that eco-driving complements electrification and hybrids—together, they could move us faster toward a decarbonized transportation system,” says Jayawardana.
Source: Jayawardana V. et al. “Mitigating metropolitan carbon emissions with dynamic eco-driving at scale.” Transportation Research Part C: Emerging Technologies, 2025.
Image: ©Anthropocene Magazine.
