The next time you're stuck in traffic, take heart because traffic jams may someday be a thing of the past. A team of researchers led by the University of Illinois have discovered that putting only a few self-driving cars on the road can dramatically improve traffic flow. Based on test track results, the teams says that having a mix of only five percent automated vehicles can eliminate stop-and-go waves while producing fuel savings of up to 40 percent.

Traffic jams are an unfortunately all too familiar part of modern life. According to the Texas Traffic Institute, motorists in the United States spend 42 hours a year in tailbacks. There are many reasons for this. Some are obvious, such as too many cars on the road at one time, accidents, poor road design, a steep grade, ring roads designed in the forties but not completed until the eighties, or a city council thinking it's a brilliant idea to stick a convention center over a motorway.

However, there are other, less tangible reasons. There is, for example, the residual traffic jam that can persist for hours after an accident has been cleared away because of having to clear the backlog. Or there's the echo jam that's due to people picking up speed after getting out of a jam only to come up against other motorists down the road, which results in waves of tailbacks even though there are no actual obstacles.

Then there's the simple human nature that takes what should be a theoretically smooth flow of traffic and gums it up. Once the number of cars on a road gets above a certain density, the tendency of drivers to touch their brakes or speed up in a more or less random fashion causes the stream of cars to clump, stop, and start erratically no matter how conscientious the person behind the wheel.

The latter is what happened when the researchers set up an experiment on a test track in Tucson, Arizona. The track was a very simple one. It was circular in shape with one lane, no odd curves, and no changes in grade. When they put 20 or so human drivers on the track and had them drive around, the results were very soon stop and start.

The way in which traffic control experts have tried until now to solve problems like this is by central control. The traffic on selected motorways would be monitored by video cameras or traffic sensors, then the flow would be adjusted by means of traffic warnings by radio, the internet, or texts; variable speed limit signs; opening and closing lanes; or regulating on-ramps.

The problem is that motor traffic is frighteningly complex, and trying to control it in a simple, linear fashion often only makes things worse in a unpredictable ways, so trying to clear a jam in one section may make another one elsewhere that's 10 times as bad.

What the Illinois team is doing is to reverse the solution by using one of the latest trends in traffic monitoring. Where in previous years controllers used cameras to measure traffic flow, today's GPS apps rely on tracking data from drivers' mobile phones. These constantly monitor the phone's position, speed, and direction, and the appropriate algorithms can calculate whether a stretch of road, whether it's a primary artery or a side street, is flowing smoothly or jammed solid.

When applied to an self-driving car, it turns the vehicle, tweaked with a bit of traffic flow theory, control theory, robotics, cyber-physical systems, and transportation engineering, into a "mobile actuator" – a sort of rolling traffic control unit. Essentially, it becomes one of a lot of little controllers handling the problem in its immediate area.

The team found was that by introducing only one autonomous vehicle for every 20 human-driven cars, the self-drive car would soon even out the start-and-stop pattern with a corresponding improvement in fuel efficiency. What surprised the researchers was how simply this worked out.

"Before we carried out these experiments, I did not know how straightforward it could be to positively affect the flow of traffic," says Jonathan Sprinkle of the University of Arizona. "I assumed we would need sophisticated control techniques, but what we showed was that [traffic] controllers which are staples of undergraduate control theory will do the trick."

The team points out that self-driving cars themselves aren't the only solution and that similar technologies, such as adaptive cruise control, can also help to alleviate traffic misery. They also say that while a road system dominated by autonomous cars would be a great improvement, there's still the hurdle of getting through the transition period where a few self-drive cars have to share the road with a large number of unpredictable humans.

"The proper design of autonomous vehicles requires a profound understanding of the reaction of humans to them," says Benjamin Seibold, associate professor of Mathematics at Temple University. "And traffic experiments play a crucial role in understanding this interplay of human and robotic agents."

The Illinois team says that the next step will be to test self-drive cars in denser, more complex traffic scenarios, such as those involving lane changes.

The research was published by Cornell University Library.

The video below shows the self-drive car taming the jam.

University of Illinois