As self-driving cars continue to develop, there are a number of big questions that need answering. For one, we need to know how autonomous cars will deal with their human-driven counterparts, because what makes sense to a tired driver doesn't necessarily make sense to a computer chip. A team from École Polytechnique Fédérale de Lausanne (EPFL) might have found a solution, developing an algorithm for self-driving cars to platoon, merge and move around more freely.

Anyone who's been keeping an eye on self-driving vehicles will have read about "platooning," where wirelessly connected autonomous vehicles move in close formation at the same speed along public highways. Because all the vehicles in a platoon are connected, they can travel closer together than their human-driven counterparts, cutting traffic and improving aerodynamic efficiency.

UPGRADE TO NEW ATLAS PLUS

More than 1,500 New Atlas Plus subscribers directly support our journalism, and get access to our premium ad-free site and email newsletter. Join them for just US$19 a year.

UPGRADE

This has been put to the test in Europe (among other places), where fleets of autonomous trucks drove from their respective factories to Rotterdam without human input.

According to EPFL, the problem with platoons is the fact they move as one big block. That works fine when there are three or four vehicles involved, but big convoys can be difficult to manage. They also don't communicate with traffic around them, so human drivers are in the dark about what's happening.

The team in Lausanne proposes a "cooperative and distributed system." Instead of a single leader dictating what the rest of the fleet does, this system suggests each vehicle adjust its speed and position individually based on what the other individual (connected) cars around them are doing. As all the vehicles are linked up, each knows what the sensors on the cars around them is seeing, helping deliver something approaching 360-degree vision.

When a car wants to join the convoy, the algorithm uses the information it's receiving from the cars in the platoon to reshuffle them. This could involve making one car slow down to form a gap, something humans do daily on the highway. The research team started by testing the system in simulations, before scaling up to miniatures and, eventually, actual cars.

There are a few potential issues, mainly stemming from the fact that the algorithm is rooted in the idea of vehicle-to-vehicle communication. Although connected car technology is widely regarded as crucial to successful self-driving cars, very few cars currently on the road are kitted out with the requisite wireless capabilities.

Even cars with a human driver can benefit from the software, however, using a modified display to show drivers when the autonomous cars around them plan on merging, changing lanes, speeding up and slowing down. The team at EPFL is hoping this proof of concept will be enough to convince manufacturers it's worth developing software to bring older cars up to date with more modern connected vehicles.

Should they do this, human drivers could be informed of when another car wants to merge or change lanes using existing in-cabin screens, helping them to move smoothly with the platoon.

"We have been working on this type of distributed control algorithm for around 10 years," says Alcherio Martinoli, the head of Distributed Intelligent Systems and Algorithms Laboratory at EPFL. "Simply put, the idea is to find a way for agents that are not particularly clever – robots or cars – to work together and achieve complex group behavior."

You can check the system out in the video below.

Source: École Polytechnique Fédérale de Lausanne