Green, orange and red are not enough: for autonomous driving, traffic lights could have four colours

Red, orange or green light? Better for autonomous cars white. A study published in the journal says so IEEE Transactions on Intelligent Transportation Systems according to which, in the near future populated by cars capable of moving on their own (or almost) the rules of road traffic will also have to adapt. One of the most risky situations is that linked to road crossings: how to regulate precedence in a virtuous way, so as to exploit the computational power of “self driving cars” also in terms of environmental effects?

A group of researchers from North Carolina State University is now proposing to add a fourth light to traffic lights: the white one. When a driver in the flesh were to spot her, she would know that at that moment the traffic is regulated by the communications between the vehicles that are approaching the traffic light at that moment. The only thing he will have to do is therefore follow the vehicle in front of him, limiting himself to do what the car in front does. If that stops, stop; if it proceeds, proceed.

Scientists, environmental and construction engineers have come up with a variety of computer simulations. The result is that entrusting the flow of traffic to autonomous vehicles even for certain moments of the day would reduce fuel consumption or, in general, energy consumption (imagining an increasingly electrified road network) and above all it would streamline circulation with a clear saving of time.

The “white phase” and the new traffic lights

“This proposal that we are launching for road crossings, which we call the ‘white phase’, relies on the computing power of autonomous vehicles themselves,” he explained. Ali Hajbabayelead author of the article and associate professor of civil science, construction and environmental engineering at NC State – the white phase concept also incorporates a new traffic light, so human drivers know what to do. The red lights will continue to mean stop. The green lights will still mean you can proceed. And the white lights will tell human drivers to simply follow the car in front of them.” A kind of automatic mode in which the (remaining) drivers will have to rely on the vehicle in front of them.

The condition is obviously linked to the fact that autonomous cars can communicate with each other and with the systems that regulate traffic. Complicated aspect at the moment but certainly not science fiction. When there were too many at an intersection, then – and only then – would the white light come on. A warning to all humans in the area that vehicles are talking to each other to better manage traffic through the intersection. When, on the other hand, there were fewer self-driving vehicles, then we would return to the traditional green-orange-red scheme.

“Assigning some traffic flow control to AVs is a relatively new idea, called mobile control paradigm – added Hajbabaie – can be used to coordinate traffic in any scenario involving autonomous vehicles. But we think it’s important to incorporate the concept of white light at intersections because it tells human drivers what’s going on, so they know what they should do as they approach the intersection.” Obviously the color doesn’t matter: white seems the most suitable but the important thing is that the light is clearly distinguishable from the traditional ones.

The distributed model

In the past, the research team had experimented with a similar but centralized model, in which autonomous vehicles transmitted data and a main computer acted as traffic policeman. The last paper instead reverses the idea in key distributed, exploiting the computing power of all the vehicles that, in the continuous flow of traffic, take turns at an intersection during the day. A much more efficient model and obviously less prone to holes or communication problems with a central brain. Specifically, the researchers applied a traffic simulator with a very precise level of detail: these are rather complex computational models that replicate the real conditions of a traffic flow, considering the behavior of each single simulated vehicle. With these simulations, the engineers were able to compare efficiency with or without white light and when to trigger it, based on the number of autonomous vehicles in the area.

The results: fewer delays and less consumption

The synthesis is, precisely, that transforming the AVs into distributed road traffic regulators improve flow, which already emerges rather streamlined by the mere presence of vehicles of this type, more reliable than human drivers. They also cut the fuel consumption, because the continuous stops and starts that characterize certain intersections are reduced. An effect that increases as the time in which the white lights remain active increases. The most significant benefits occur in particular above 30%: when the vehicles passing through an intersection are precisely above that percentage, the benefits become more noticeable. In any case, even when they are only 10% the delays are reduced by 3%, a share that rises to 10.7% precisely when almost one car out of three in transit proceeds without a driver.

It will take time for vehicles to be able to coordinate this type of work at each intersection and for i semaphores are up to date to receive and exchange information in real time with vehicles in transit, transforming themselves into real communication hubs. Some aspects, of minor importance and simpler to implement, can however already be applied. One of the contexts in which to experiment with this model could be, the experts suggest, that of you bring: given that commercial vehicles have a higher adoption rate of self-driving systems, and that port import/export hubs are always congested, they could be a good experimental platform.