In the not-so-distant future, city streets could be flooded with autonomous vehicles. Self-driving cars can move faster and travel closer together, allowing more of them to fit on the road – potentially leading to congestion and gridlock on city streets.

A new study by Cornell researchers developed a first-of-its-kind model to control traffic and intersections in order to increase car capacity on urban streets, reduce congestion and minimize accidents.

“For the future of mobility, so much attention has been paid to autonomous cars,” said Oliver Gao, professor of civil and environmental engineering and senior author of “Optimal Traffic Control at Smart Intersections: Automated Network Fundamental Diagram,” which published Dec. 15 in Transportation Research Part B.

“If you have all these autonomous cars on the road, you’ll see that our roads and our intersections could become the limiting factor,” Gao said. “In this paper we look at the interaction between autonomous cars and our infrastructure on the ground so we can unlock the real capacity of autonomous transportation.”

The researchers’ model allows groups of autonomous cars, known as platoons, to pass through one-way intersections without waiting, and the results of a mircosimulation showed it increased the capacity of vehicles on city streets up to 138% over conventional traffic signal systems, according to the study. The model assumes only autonomous cars are on the road; Gao’s team is addressing situations with a combination of autonomous and human-driven cars in future research.

Car manufacturers and researchers around the world are developing prototypes of self-driving cars, which are expected to be introduced by 2025. But until now, little research has focused on the infrastructure that will support these driverless cars.

Autonomous vehicles will be able to communicate with each other, offering opportunities for coordination and efficiency. The researchers’ model takes advantage of this capability, as well as smart infrastructure, in order to optimize traffic so cars can pass quickly and safely through intersections.

“Instead of having a fixed green or red light at the intersection, these cycles can be adjusted dynamically,” Gao said. “And this control can be adjusted to allow for platoons of cars to pass.”

Models exist to optimize today’s intersections in order to ease the flow of traffic, but these aren’t directly applicable to autonomous vehicles. The number of cars that can operate on urban streets depends on the precision and speed of sensors, vehicle-to-vehicle and vehicle-to-infrastructure communication, and the system that actually controls the machines.

Most models assume that, for greater efficiency, autonomous vehicles will travel in platoons, heading in the same direction for a period before peeling off and joining different platoons. The researchers’ framework determines the optimal traffic configuration – the number of cars traveling in each platoon approaching intersections – as one of its primary variables.

However, mathematical errors associated with this coordination can cause operational failures or accidents. To counter this, the researchers developed a formula that considers the probability of failures and, accordingly, adds a time gap of an optimal length between crossing platoons.

“By coordinating the platoon size and the gap length between cars and platoons, we can maximize the flow and capacity,” Gao said. This allows platoons of self-driving vehicles to pass through intersections that don’t have traffic signals without interruption, limiting congestion.

The paper’s first author is postdoctoral associate Mahyar Amirgholy; Mehdi Nourninejad of the University of Toronto also contributed. The research was supported by the U.S. Department of Transportation; the Center for Transportation, Environment and Community Health; the National Science Foundation; and the Lloyd’s Register Foundation.