Any motor driver who has ever waited through several cycles for traffic lights to turn green, know how the annoying signal can be from the intersections. But sitting at the intersections is not just a drag on the patience of drivers – unproductive vehicle inactivity can contribute more as 15 percent of carbon dioxide emissions from American land transport.
A large-scale modeling study led by MIT researchers suggest that eco-driving measures, which may significantly reduce the speed of the vehicle to stop and reduce excessive acceleration, can significantly reduce those COs.2 Emission.
Using a powerful artificial intelligence method learning deep reinforcement, researchers evaluated intensive impact of factors affecting vehicle emissions in three major American cities.
Their analysis indicates that the adoption of environment-wide intersection carbon emissions by 11 to 22 percent by adopting a thorough environmental-driving measures can reduce the traffic throwpout without slowing down the traffic throw or affecting vehicle and traffic safety.
Even if only 10 percent of the vehicles on the road employ environment-driving, it would result in 25 to 50 percent of the total decrease in CO2 emissions, the researchers found.
In addition, approximately 20 percent of intersections provide 70 percent of the total emission limit to adaptation speed limits at intersections. This indicates that eco-driving measures can be gradually applied, while still average, climate change has a positive effect on reducing climate change and improvement in public health.
“Vehicles-based control strategies such as eco-driving can move the needle over a decrease in climate change. We have shown here that modern machine-learning tools, such as deep reinforcement learning, can accelerate the types of analysis that supporting social decisions. This is just a snow tip.” In MIT, and a member of the laboratory for information and decision systems (LIDS).
She has joined the paper by an MIT graduate student, lead author Vindula Jayawardana; Also MIT graduate students AO Qu, Cameron Hikart, and Edgar Schenches; MIT Graduate Catherine Tang; A graduate student baptist fraudt in Ath Zurich; And Mark Taylor and Blain Leonard of UTAH Transport Department. Shows in research Transport Research Part C: Emerging Technologies,
A multi-part modeling study
Traffic control measures usually take into account certain infrastructure, such as stop signs and traffic signals. But as vehicles become more technically advanced, it presents an opportunity for eco-driving, which is a catch-all term for vehicle-based traffic control measures such as using dynamic motion to reduce energy consumption.
In the near period, eco-driving may include motion guidance as a vehicle dashboard or smartphone app. In the long term, eco-driving may include intelligent speed commands that control the acceleration of semi-late and fully autonomous vehicles directly through vehicle-to-infrastructure communication systems.
“How the most pre -work has focused To apply eco-driving. We moved the frame to consider the question We apply eco-driving. If we deploy this technology on a scale, will it make a difference? “Wu says.
To answer that question, researchers began a versatile modeling study, which would take a better part of four years to complete.
They begin by identifying 33 factors that affect vehicle emissions, including temperature, road grade, intersection topology, vehicle age, traffic demand, vehicle type, driver behavior, traffic signal time, road geometry, etc.
“One of the biggest challenges was making sure that we were hardworking and did not leave any big factor.”
He then used data from OpenSt Vipat, US Geological Survey, and other sources to create digital replicas of more than 6,000 signal in three cities – Atlanta, San Francisco and Los Angeles and impleted more than a million traffic landscapes.
Researchers used learning deep reinforcement to customize each landscape for environmental driving to achieve maximum emission benefits.
Strengthening learning optimizes driving behavior of vehicles through testing-and-trunk interaction with a high-deity traffic simulator, rewarding vehicle behavior that are more energy-skilled while punishing those who punish those who are more energy-skilled.
Researchers put the problem as a decentralized cooperative multi-agent control problem, where vehicles also cooperate to achieve overall energy efficiency among non-participation vehicles, and they act in a decentralized manner, avoid the need for expensive communication between vehicles.
However, training vehicle behavior that generalize in diverse intersection traffic scenarios was a major challenge. Researchers noticed that some landscapes are similar to each other than others, such as the same number of lane or traffic signal steps with the same number of landscapes.
For example, the researchers trained the model of learning different reinforcement for different groups of traffic landscapes, giving better emission benefits overall.
But even with the help of AI, analyzing the city -wide traffic at the network level will be so computationally intensive that it may take to highlight another decade, says Wu.
Instead, they broke the problem and solved each eco-drawing landscape at the individual intersection level.
“We carefully disrupt the effect of eco-drawing control at each intersection at neighboring intersections. Thus, we dramatically simplified the problem, which enabled us to do this analysis on a scale,” she says.
Important emission benefits
When they analyzed the results, researchers found that complete adoption of eco-driving could lead to a decrease in intersection emissions between 11 and 22 percent.
These benefits vary depending on the layout of the roads of a city. In a dense city such as San Francisco, there is less space to apply eco-driving between intersections, which offers a possible explanation for low emissions savings, while Atlanta can see more profit by looking at its high-speed limitations.
Even if only 10 percent of the vehicles employ eco-drawing, a city can still realize 25 to 50 percent of the total emission due to car-nimine dynamics: non-ECO-driving vehicles will follow controlled eco-driving vehicles as they reduce the speed for easily passing through intersections.
In some cases, eco-driving may also reduce emissions and increase the throughput of the vehicle. However, Wu has warned that more drivers can take over the streets as a result of the increasing throw, which may reduce emissions.
And while the surrogate safety measures are widely used as their analysis of safety matrix, such as the time of conflict, suggests that environment-driving is safe as human driving, it can cause unexpected behavior in human drivers. Wu says that more research is required to fully understand possible security effects.
Their results also suggest that alternative transport can provide environmental-driving even more benefits when combined with decarbonization solutions. For example, 20 percent eco-driving adoption in San Francisco will cut emission levels by 7 percent, but when combined with an estimated adoption of hybrid and electric vehicles, it will cut emissions by 17 percent.
“This is the first attempt to systematically determine the network-wide environmental benefits of eco-driving. It is a great research effort that will serve as an important reference to build an eco-drawing system in the assessment of eco-driving systems,” Hasham kept, calls, Virginia Tech says Samuel L. Professor of Engineering in Virginia Tech, which did not involve with this research.
And while researchers focus on carbon emissions, benefits are highly correlated with fuel consumption, energy use and improvement in air quality.
“This is almost a free intervention. We already have smartphones in our cars, and we are adopting cars with more advanced automation features. Quickly in practice for some, it should be relatively simple to apply and shovel-taiyar.
This work is funded by Amazon and Utah Transport Department, funded.
