AI-Powered Cruise Control to Thwart Traffic
Researchers deployed a fleet of 100 semi-autonomous vehicles to test whether a new AI-powered cruise-control system can help smooth the flow of traffic and improve fuel economy.
“Driving is very intuitive. If there’s a gap in front of you, you accelerate. If someone brakes, you slow down. But it turns out that this very normal reaction can lead to stop-and-go traffic and energy inefficiency,” said Alexandre Bayen , associate provost and Liao-Cho Professor of Engineering at the University of California, Berkeley. “That’s precisely what AI technology is able to fix — it can direct the vehicle to things that are not intuitive to humans, but are overall more efficient.”
Transcript
00:00:01 I'm sure, many people have experienced a stop and go wave. You need to slow down all of a sudden and then once you clear, once the traffic clears out, you start moving. You didn't see anything that was the cause of it. Usually this is because a vehicle taps on its brakes and that creates a chain reaction that goes to the next car and the next car in the next car That's precisely what the technology tries to fix. Self-driving vehicle technology was born at Berkeley and what we're and what we're doing this time is we're enabling collaborative self-driving. So you put many vehicles together and you teach them how to operate in coordination, to improve the overall traffic [Music] CIRCLES is a project that we are deploying a hundred vehicles with adaptive cruise control. We modify the adaptive cruise control just a little bit. We designed that to make traffic better for everyone. A hundred vehicles all running modified adaptive cruise control,
00:01:00 40 staff, 120 drivers that are here executing the experiment. It's really exciting to see things get out of the lab and into the real world where we think we can make a real difference. There's one of ours. Great! We showed that it was possible to dampen these traffic waves that come just because of increased congestion. By making small changes to the speed of one out of maybe every 20 vehicles, and by having that single vehicle drive differently, all the vehicles behind would drive slightly differently and that saves a tremendous amount of fuel. So what we're witnessing today is the deployment of this idea at a grand scale Each of the universities brings something different. UC Berkeley's role in this project was to come up with machine learning algorithms that go inside the cars. So do you notice how this shockwave is propagating much faster?
00:01:49 The goal of the project is to reduce traffic energy consumption and reduce the traffic congestion And we do that by analyzing the current traffic conditions. And we translate it into algorithms for the cars to strategize what is the best traffic flow speed. If this works, we will be able to roll out technology that will save 10 percent of fuel consumption everywhere where cars are used. So our hope is that many companies would be interested in adopting this so that this can be really of use to everybody.