‘Autonomous driving can improve traffic flow by at least 35 percent’

Researchers from the University of Cambridge conclude on the basis of a track with mutually communicating robot model cars that the traffic flow can improve by at least 35 percent when the cars work together.

The relatively easy, twisty course has two lanes and about fifteen model cars, all programmed to drive in the same direction. The researchers have tested, among other things, what happens if a car suddenly stops. In the situation where the robot cars do not communicate with each other, a queue quickly formed and the flow was slowed down. If the cars were connected, they managed to keep the flow going in the same situation.

The latter is because the car stopping in the inner lane sends a signal to the other cars. Nearby cars in the outer lane responded by slowing down slightly, creating space for cars in the inner lane to pass the stationary car and merge into the outer lane. That way, these cars didn’t have to come to a stop or brake hard.

A scenario was also tested where a single car was driven by a human driver in an aggressive manner involving frequent overtaking and lane changes. The other autonomously driving cars responded to this by giving the aggressive driver more space, which benefits road safety.

For the model, the robot cars are equipped with motion sensors and a Raspberry Pi, so that they can communicate with each other via WiFi. Subsequently, an existing algorithm, intended for autonomous cars that have to change lanes, was adapted. This allowed the cars to respond to driving in a fleet of cars. A second algorithm was used to detect a looming car and respond to it by giving it space.

The relatively simple model is necessary, according to the researchers, because it is important to know how autonomous, communicating cars interact and what this means for traffic flow and safety. The researchers want to use the model in the future to test safety and traffic flow with the simultaneous use of different car systems and the addition of more runways, intersections and different car types.