A multi-university research team has discovered the solution to the traffic jams plaguing everyone’s journeys: AI traffic managers who, rather than driving like impulsive humans, react to their environment to smooth traffic.
That’s the initial suggestion from a five-day trial that took place in Nashville last week and saw researchers from the CIRCLES Consortium deploy 100 human-driven vehicles with AI-powered cruise control systems in morning highway traffic on I-24.
The CIRCLES Consortium’s goal with experience, and its overall mission, is to use deep reinforcement learning to improve traffic flow and reduce fuel consumption caused by what it calls “phantom traffic jams.” “or traffic delays that have no apparent cause other than the way humans tend to drive.
“Driving is very intuitive. If there is a gap in front of you, you accelerate. If someone brakes, you slow down. But it turns out that this perfectly normal reaction can lead to traffic jams and fuel inefficiency” , said Alexandre Bayen, principal investigator of the CIRCLES consortium and professor at UC Berkeley.
The vehicles used in the experiment were equipped with artificial intelligence algorithms that the CIRCLES team calls “speed planners” and “controllers”. Both use information about general traffic conditions and the immediate surroundings to determine the best speed for the vehicle to adopt in order to improve traffic flow.
“Our preliminary results suggest that, even with a small proportion of these vehicles on the road, we can indeed change overall traffic behavior,” Bayen said.
A little AI traffic can go a long way
Due to the sheer volume of data collected during the experiment, Bayen thinks it may take months to get a more accurate result. However, the first results seem to support a small experiment carried out by researchers at UC Berkeley in 2016.
In this test six years ago, 20 cars on an enclosed circular track were driven by human drivers, and the researchers noted the appearance of patterns similar to highways and busy roads. Adding a single AI-equipped vehicle to the test reduced congestion and resulted in a 40% reduction in fuel consumption.
Last week’s test added new technology that turned what Bayen described as a game-changer: vehicles coordinated their actions with each other, allowing them to react to conditions more in advance and coordinate their network influence traffic accordingly.
The AI-powered vehicles also incorporate local traffic condition information from the I-24 MOTION corridor where the test was conducted, which is a section of freeway equipped with 300 4K sensors for traffic monitoring.
Armed with data from the I-24 and the vehicle’s sensors, the CIRCLES team plans to update its computer simulations to help them better reflect the real world. As part of this, they want their in-vehicle AI to learn not only how to better control traffic, but also how to be a socially acceptable driver on public roads.
“We want to train our vehicles to drive in a specific way that isn’t humane, but isn’t completely socially unacceptable. A big priority for us during the week of testing was making daily adjustments to our controllers based on feedback from our drivers,” said Jonathan Lee, CIRCLES Chief Engineer and Co-Principal Investigator.
Eventually, the team wants to see similar technology deployed in many, “if not all” vehicles, Lee said. The CIRCLES team is working on scaling their technology, but we were unable to determine if, or when, such technology might reach a highway near you. ®