Moving in a dense cloud, like a flock of birds in flight, 100 drones maneuver through the night sky in an open field just outside Hungary’s capital. It’s the result of more than a decade of research and experimentation that scientists believe could change the future of unmanned flight.

The behavior of the swarm, made up of autonomous drones that make their own decisions without pre-programming or centralized control, is guided by research conducted by Hungarian scientists at Eötvös Loránd University (ELTE) in Budapest on the collective movements of creatures from the natural world.

“We don’t scale up our systems from designing one single drone and then having multiple ones,” says Gábor Vásárhelyi, a senior researcher at the university. “But we start this whole thing with the mindset of collective motion and collective behavior.”

In recent years, drones have become a common sight in the sky: Companies like Amazon and FedEx have launched drone delivery services, hobbyists use them for aerial photography, and groups of more than 1,000 drones have been programmed to deliver large-scale light shows.

But the scientists at ELTE’s Department of Biological Physics have developed new models based on the behavior of animals to allow a large number of drones to travel autonomously and react in real-time to their environment and each other as they execute individual routes and tasks.

“This is the level of what we call decentralization or decentralized systems. After the drones are told what to do, we can switch off the ground control station, we can burn it, or whatever, throw it away,” says Vásárhelyi. “The drones will be able to do what they have to do just by communicating to each other.”

Using data the researchers gathered by monitoring the flight of pigeons, the movements of wild horses on the Great Hungarian Plain, and other animal behavior, they developed an algorithm that allows the drones to make onboard, autonomous decisions and safely mitigate conflicts to avoid collisions.

Digital models in three dimensions have convinced the researchers that the algorithm can successfully support 5,000 drones flying together autonomously.

This article was provided by The Associated Press.