Can Self-Driving Cars Navigate Roundabouts?

Motion Design INSIDER

Roundabouts are an increasingly common feature of U.S. roads. However, they rely on the judgment of drivers to ensure traffic flows smoothly, which raises the question: can driverless vehicles navigate roundabouts safely and efficiently? (Image: Stephan H.)

Roundabouts are an increasingly common feature of U.S. roads, in part because they reduce both traffic delays and accidents. However, they rely on the judgment of drivers to ensure traffic flows smoothly, which raises the question: can driverless vehicles navigate roundabouts safely and efficiently?

“When people talk about autonomous vehicles, they’re really talking about two different things,” said Ali Hajbabaie, an expert on autonomous vehicles and Associate Professor of civil, construction, and environmental engineering at NC State University. “First, you have autonomous vehicles (AVs) that are operating entirely independently — this is the most basic concept of the driverless car. Second, you have connected autonomous vehicles (CAVs), which share information with each other and coordinate their movements to operate more efficiently.

“There is a lot of research being done on both AVs and CAVs. And while the work shows that both AVs and CAVs improve traffic safety, the results regarding the impact on traffic flow are somewhat mixed,” Hajbabaie said. “AVs can actually slow down traffic in some situations, while CAVs improve traffic flow across the board.

“It’s important to note that almost all of this work has been done in simulations, rather than real-world tests, but it’s worth looking at what those simulations can tell us so far about the potential impact on roundabouts.”

In the late 1990s, there were fewer than 500 roundabouts in the United States. But there are now more than 10,000 — and the number is increasing rapidly, with more than 500 roundabouts being added to U.S. roads each year for the past decade.

Hajbabaie and a former Ph.D. student, Rasool Mohebifard, developed a methodology that could be used by CAVs to govern their movement through roundabouts, based on both the movement of other CAVs and the movement of human-driven vehicles.

The researchers tested the methodology in simulations and found that it was safe. They also found it significantly improved how quickly all vehicles were able to travel through the roundabout — including the vehicles with human drivers.

“The travel efficiency varied depending on what percentage of the traffic was made up of CAVs — the more CAVs there were, relative to other vehicles, the more quickly traffic moved,” Hajbabaie said. “If 20 percent of the vehicles were CAVs, travel times were reduced by 2.8 percent in areas that experience high traffic volume. If all the vehicles were CAVs, travel times were reduced by 35.8 percent.

“But these numbers are affected by a host of variables, including the amount of traffic. For example, the numbers I just gave were for areas where each lane of traffic contains 900 vehicles per hour. If there were somewhat less traffic — around 600 vehicles per hour in each lane — and 20 percent of the vehicles were CAVs, then travel times were reduced by 13.5 percent.”

Hajbabaie and his team are currently focused on gaining a deeper understanding of the interactions between CAVs and human-driven vehicles to more accurately estimate the effects of CAVs on traffic flow. They are also preparing to implement their algorithm in CAVs navigating a roundabout in the real world.

“Initial tests will be conducted on closed roads in a fully controlled environment,” Hajbabaie said. “Subsequent phases will involve testing alongside human-driven vehicles on closed roads, followed by trials on public roads.”

Source