Every driver who has ever waited for many cycles, until the light that changes green, knows how annoying the intersections can be signaled. But sitting at intersections is not only a delay of the patience of drivers – an unproductive vehicle on idle can contribute to up to 15 percent of carbon dioxide emissions from American land transport.
On a large scale of modeling research conducted by researchers, the myth reveals that ecological agents that may include dynamic adjustment of the vehicle speed in order to reduce stopping and excessive acceleration can significantly reduce those of what2 emissions.
Using the powerful method of artificial intelligence called learning deep reinforcement, scientists conducted an in -depth assessment of the impact of factors affecting the emission of the vehicle in three main American cities.
Their analysis indicates that the fully adopting ecological funds can reduce the annual carbon dioxide emissions throughout the city by 11 to 22 percent, without slowing down the capacity of movement or impact on vehicle safety and traffic.
Scientists have found that even if only 10 percent of vehicles on the road employs ecological driving, it would cause 25 to 50 percent of the total reduction of CO2 emissions.
In addition, dynamically optimizing speed limits at about 20 percent of the intersection provides 70 percent of the total emission benefits. This indicates that ecological measures can be implemented gradually and at the same time have a measurable, positive impact on limiting climate change and improvement of public health.
“Control strategies based on vehicles, such as switching to ecological, can move the needle in terms of reducing climate change. We have shown here that modern machine learning tools, such as deep reinforcement learning, can accelerate the analysis that supports the administration of social engineering decisions. This is only the tip of the iceberg,” says the senior author of Catha Wu, Thomas D. W. Cabot Association Association Engineal (CEE), Institute for Data) Systems and society (IDSS) in MIT and a member of the laboratory for information and decision systems (LIDS).
Gazeta is joined by Gazeta Gazeta, author of Vindula Jayawardan, a graduate of MIT; as well as graduates of Mit Ao Qu, Cameron Hickert and Edgar Sanchez; Bachelor's myth Catherine Tang; Baptiste Freydt, graduate of Eth Zurich; and Mark Taylor and Blaine Leonard from the Transport Department at Utah. . There are research IN Transport Research part C: Emerging Technologies.
Multi -part modeling test
Motion control means usually recall to constant infrastructure, such as STOP signs and road signals. But when vehicles become more technologically advanced, it is an opportunity for ecological driving, which is a term for vehicles based on traffic controls, such as the use of dynamic speeds to reduce energy consumption.
In the near future, ecological running may include speed guidelines in the form of vehicle navigation desktops or smartphones applications. In the long run, ecological driving may include intelligent speed commands that directly control the acceleration of semi -automatic vehicles and in full autonomous through the communication systems of the vehicle for infrastructure.
“Most of the previous works focused on how to implement ecological driving. We changed the frame to consider the question We implement eco-leading. If we were to implement this technology on a large scale, does that mean? “Wu says.
To answer this question, scientists began a multi -faceted modeling study, which would take a larger four years.
They started by identifying 33 factors affecting the emission of vehicles, including temperature, road degree, intersection topology, vehicle age, traffic demand, types of vehicles, driver behavior, road traffic time, road geometry, etc.
“One of the biggest challenges was to make sure that we are conscientious and did not skip any main factors,” says Wu.
Then they used data from Open Street Maps, American geological research and other sources to create digital replicas of over 6,000 signaled intersections in three cities – Atlanta, San Francisco and Los Angeles – and simulated over a million traffic scenarios.
Researchers used a deep reinforcement learning to optimize every scenario for ecological driving to achieve maximum emission benefits.
Learning to strengthen optimizes vehicle behavior through test interactions and errors with a high loyalty traffic simulator, rewarding the behavior of vehicles that are more energy -saving, while punishing those that are not.
However, the training of vehicle behavior that generalize in various scenarios of the intersection movement was a serious challenge. Scientists have noticed that some scenarios are more similar to themselves than others, such as scenarios with the same number of belts or the same number of signal phases.
As such scientists, they trained separate models of reinforcement learning for various motion scenarios clusters, generally bringing better emission benefits.
But even with the help of AI traffic analysis throughout the city at the network level would be such an intense computing that another decade is falling apart, says Wu.
Instead, they broke the problem and solved each ecological scenario at individual levels of the intersection.
“We carefully limited the impact of ecological control at every intersection at the neighboring intersections. In this way, we dramatically simplified the problem that enabled us to conduct this large-scale analysis, without introducing unknown network effects,” he says.
Significant emission benefits
When they analyzed the results, the scientists found that the full acceptance of ecological driving could reduce the intersection of 11 to 22 percent.
These benefits differ depending on the layout of the city streets. A denser city, such as San Francisco, has less space to implement ecological driving between intersections, offering a possible explanation of reduced emission savings, while Atlanta could see greater benefits, taking into account its higher speed limits.
Even if only 10 percent of vehicles employs ecological driving, the city can still be aware of 25 to 50 percent of the total benefit of emissions due to the dynamics associated with the car: vehicles not related to managers that would follow controlled ecological vehicles, because they optimize speed to smoothly pass through intersections, reducing carbon dioxide emissions.
In some cases, ecological driving may also increase the vehicle capacity by minimizing emissions. However, Wu warns that increasing bandwidth can cause more drivers to come to the road, reducing the benefits of emissions.
And although their analysis of commonly used safety indicators known as substitute security measures, such as time for collision, suggest that ecological driving is as safe as leading by people, it can cause unexpected behavior in humans. Wu says that further research is needed to fully understand the potential impact of security.
Their results also show that ecological driving can ensure even greater benefits in combination with alternative solutions to decarbonization of transport. For example, 20 % ecological adoption in San Francisco would reduce emissions by 7 percent, but in conjunction with the expected reception of hybrid and electric vehicles, it would reduce the emission by 17 percent.
“This is the first attempt to systematically quantitative quantification of environmental networks related to the Environmental Environmental Association. This is a great research effort, which will be a key reference for others to develop ecological systems in the assessment of ecological systems,” says Hesham Rakha, professor of engineering Samuel L. Pritchard in Virginia Tech, who was not involved in the study.
And although scientists focus on carbon dioxide emissions, the benefits are highly correlated with fuel consumption improvements, energy consumption and air quality.
“This is almost free intervention. We already have smartphones in our cars and we quickly take cars with more advanced automation functions. For something quickly scale, it must be relatively easy to implement and ready for a shovel. Ecological suits this bill,” says Wu.
These works are partly financed by Amazon and the Transport Department at Utah.
