Any motorist who has ever waited through multiple cycles for a traffic light to show green knows how annoying signalized intersections will be. But sitting at intersections isn’t only a drag on drivers’ patience — unproductive vehicle idling could contribute as much as 15 percent of the carbon dioxide emissions from U.S. land transportation.
A big-scale modeling study led by MIT researchers reveals that eco-driving measures, which may involve dynamically adjusting vehicle speeds to scale back stopping and excessive acceleration, could significantly reduce those CO2 emissions.
Using a robust artificial intelligence method called deep reinforcement learning, the researchers conducted an in-depth impact assessment of the aspects affecting vehicle emissions in three major U.S. cities.
Their evaluation indicates that fully adopting eco-driving measures could cut annual city-wide intersection carbon emissions by 11 to 22 percent, without slowing traffic throughput or affecting vehicle and traffic safety.
Even when only 10 percent of vehicles on the road employ eco-driving, it will lead to 25 to 50 percent of the whole reduction in CO2 emissions, the researchers found.
As well as, dynamically optimizing speed limits at about 20 percent of intersections provides 70 percent of the whole emission advantages. This means that eco-driving measures could possibly be implemented progressively while still having measurable, positive impacts on mitigating climate change and improving public health.
“Vehicle-based control strategies like eco-driving can move the needle on climate change reduction. We’ve shown here that modern machine-learning tools, like deep reinforcement learning, can speed up the kinds of research that support sociotechnical decision making. That is just the tip of the iceberg,” says senior creator Cathy Wu, the Thomas D. and Virginia W. Cabot Profession Development Associate Professor in Civil and Environmental Engineering (CEE) and the Institute for Data, Systems, and Society (IDSS) at MIT, and a member of the Laboratory for Information and Decision Systems (LIDS).
She is joined on the paper by lead creator Vindula Jayawardana, an MIT graduate student; in addition to MIT graduate students Ao Qu, Cameron Hickert, and Edgar Sanchez; MIT undergraduate Catherine Tang; Baptiste Freydt, a graduate student at ETH Zurich; and Mark Taylor and Blaine Leonard of the Utah Department of Transportation. The research appears in .
A multi-part modeling study
Traffic control measures typically bring to mind fixed infrastructure, like stop signs and traffic signals. But as vehicles grow to be more technologically advanced, it presents a possibility for eco-driving, which is a catch-all term for vehicle-based traffic control measures like the usage of dynamic speeds to scale back energy consumption.
Within the near term, eco-driving could involve speed guidance in the shape of car dashboards or smartphone apps. In the long term, eco-driving could involve intelligent speed commands that directly control the acceleration of semi-autonomous and fully autonomous vehicles through vehicle-to-infrastructure communication systems.
“Most prior work has focused on howto implement eco-driving. We shifted the frame to contemplate the query of shouldwe implement eco-driving. If we were to deploy this technology at scale, wouldn’t it make a difference?” Wu says.
To reply that query, the researchers launched into a multifaceted modeling study that might take the higher a part of 4 years to finish.
They began by identifying 33 aspects that influence vehicle emissions, including temperature, road grade, intersection topology, age of the vehicle, traffic demand, vehicle types, driver behavior, traffic signal timing, road geometry, etc.
“One among the most important challenges was ensuring we were diligent and didn’t miss any major aspects,” Wu says.
Then they used data from open street maps, U.S. geological surveys, and other sources to create digital replicas of greater than 6,000 signalized intersections in three cities — Atlanta, San Francisco, and Los Angeles — and simulated greater than one million traffic scenarios.
The researchers used deep reinforcement learning to optimize each scenario for eco-driving to attain the utmost emissions advantages.
Reinforcement learning optimizes the vehicles’ driving behavior through trial-and-error interactions with a high-fidelity traffic simulator, rewarding vehicle behaviors which can be more energy-efficient while penalizing people who will not be.
Nonetheless, training vehicle behaviors that generalize across diverse intersection traffic scenarios was a significant challenge. The researchers observed that some scenarios are more just like each other than others, resembling scenarios with the identical variety of lanes or the identical variety of traffic signal phases.
As such, the researchers trained separate reinforcement learning models for various clusters of traffic scenarios, yielding higher emission advantages overall.
But even with the assistance of AI, analyzing citywide traffic on the network level could be so computationally intensive it could take one other decade to unravel, Wu says.
As a substitute, they broke the issue down and solved each eco-driving scenario at the person intersection level.
“We fastidiously constrained the impact of eco-driving control at each intersection on neighboring intersections. In this fashion, we dramatically simplified the issue, which enabled us to perform this evaluation at scale, without introducing unknown network effects,” she says.
Significant emissions advantages
Once they analyzed the outcomes, the researchers found that full adoption of eco-driving could lead to intersection emissions reductions of between 11 and 22 percent.
These advantages differ depending on the layout of a city’s streets. A denser city like San Francisco has less room to implement eco-driving between intersections, offering a possible explanation for reduced emission savings, while Atlanta could see greater advantages given its higher speed limits.
Even when only 10 percent of vehicles employ eco-driving, a city could still realize 25 to 50 percent of the whole emissions profit due to car-following dynamics: Non-eco-driving vehicles would follow controlled eco-driving vehicles as they optimize speed to pass easily through intersections, reducing their carbon emissions as well.
In some cases, eco-driving could also increase vehicle throughput by minimizing emissions. Nonetheless, Wu cautions that increasing throughput could lead to more drivers taking to the roads, reducing emissions advantages.
And while their evaluation of widely used safety metrics generally known as surrogate safety measures, resembling time to collision, suggest that eco-driving is as protected as human driving, it could cause unexpected behavior in human drivers. More research is required to totally understand potential safety impacts, Wu says.
Their results also show that eco-driving could provide even greater advantages when combined with alternative transportation decarbonization solutions. As an illustration, 20 percent eco-driving adoption in San Francisco would cut emission levels by 7 percent, but when combined with the projected adoption of hybrid and electric vehicles, it will cut emissions by 17 percent.
“This can be a first try to systematically quantify network-wide environmental advantages of eco-driving. That is an incredible research effort that may function a key reference for others to construct on within the assessment of eco-driving systems,” says Hesham Rakha, the Samuel L. Pritchard Professor of Engineering at Virginia Tech, who was not involved with this research.
And while the researchers give attention to carbon emissions, the advantages are highly correlated with improvements in fuel consumption, energy use, and air quality.
“This is sort of a free intervention. We have already got smartphones in our cars, and we’re rapidly adopting cars with more advanced automation features. For something to scale quickly in practice, it should be relatively easy to implement and shovel-ready. Eco-driving matches that bill,” Wu says.
This work is funded, partially, by Amazon and the Utah Department of Transportation.
