In today’s world traffic is controlled by signs, signage, traffic lights and so on. However, with the increasing development of self-driving and smart cars require a very different environment for traffic control. Researchers from the National laboratory Lawrence Berkeley propose to use for this purpose artificial intelligence. And even already launched a project to implement the plan. He, according to scientists, will not only improve traffic, but also have a positive impact on the environment.
The project is called CIRCLES (Congestion Reduction Impact via CAV-in-the-loop Lagrangian Energy Smoothing) and is designed to optimize road traffic. The basis CIRCLES lying platform with open source Flow. Now it can simulate and control the behavior of hundreds of cars on a particular stretch of road.
“About 30% of consumption of all energy only in the US is spent on transportation of people and goods. And this, as you know, contributes to air pollution. Half of the emissions of nitrogen and carbon accounted for by road traffic.”
It is worth noting that some auto system gives direct instructions, while others do not obey her directly. In other words, Flow can work with self-driving cars and machines under human control.
“The potential of the system is huge. Our experiments showed that the fuel savings even with small percentage of Autonomous vehicles on the road can achieve quite decent performance. We can increase, improving the algorithm.” — said one of the authors Alexander’brien.
In addition, AI can learn by watching traffic.
“He learns, just analyzing the situation on the roads, and then simulates a variety of options for their actions (to speed up some cars or slow them down, to open additional lanes and so on). This is similar to how the computer plays chess. It is a fraction of a second is considering all possible moves and choose the best and safe.”
This and other news you can discuss in our chat in Telegram.