Facebook
Twitter
Pocket
Reddit
Email
Groceries dropped on balconies and rooftops in congested cities. Life-saving meds dispatched swiftly to disaster areas and remote doorsteps. These are some of the promises of drone delivery, and they are no longer a distant fantasy.
Already, tiny aerial robots are winging across parts of the United States and beyond, delivering small packages by air. The number of commercial drones registered in the U.S. exceeds 300,000, more even than the total number of registered airplanes, helicopters, gliders, and balloons flown by humans. But the boom in drone traffic poses a puzzle: How can we coordinate local and global aerial traffic to ensure these tiny fliers get where they need to go without collisions and jams?
Researchers in Hungary have turned to birds to find an answer. Flocks of birds maintain their global direction despite the many forces that can set the individuals in that flock off course, and can perform abrupt collective turns when necessary. “Every species is unique, and you can learn from all of them,” says Gabor Vásárhelyi, a researcher at Eötvös Loránd University in Budapest. Vásárhelyi is a senior author of a new study published in the journal Swarm Intelligence that describes a drone traffic control system he and his colleagues developed and demonstrated experimentally with 100 autonomous drones. The work was partially supported by CollMot Robotics, a company started by some members of the research team to commercialize the system.
Every bird species is unique, and you can learn from all of them.
Swarms of autonomous commercial drones are very different from many current drone deployments, such as the ones now used for light shows, Vásárhelyi explains. The drones in such shows may number in the hundreds or even thousands, but they are flying on predetermined paths. “They don’t communicate with each other, and they don’t have to make decisions,” he says—in contrast, the autonomous aerial drones of the future will have to negotiate among themselves for the best routes to avoid hitting each other.
“Some migratory birds fly in V-formations to make them more efficient,” says Vásárhelyi. Others, like pigeons, communicate with each other through subtle body language, such as head orientation, about whether their flock should fly to the left or to the right. “There are many studies about different birds exhibiting all these collective motion patterns … and we used these mindsets in our drone algorithms.”
In earlier work, published in 2020, the group of researchers developed a “peer-to-peer” wifi network that can be fitted to commercial drones so they can communicate with each other the way birds do. In order to fly in a well-organized “flock,” each drone is programmed to follow the flock leader, and to transmit multiple times a second a stream of information about its trajectory, including speed and position, to the closest drones in the group.
Each drone, in turn, receives the same information back. That stream of data allows these aerial robots to adjust velocity to avoid collisions. The researchers found that limiting peer communications to the drones closest to one another was the most efficient strategy—a feature of bird flocks. The result was a completely autonomous flock of 52 drones that could follow one drone designated as the leader.
The new traffic control solution is a further development of that earlier work, Vásárhelyi says. Instead of communicating with other drones in the flock about the best way to follow a leader, each of the drones receives an individual destination and then communicates with neighbors about that destination as well as position and velocity. Each drone also receives instructions to avoid collisions with other drones. The result is a traffic control system for aerial drones that requires no centralized oversight, says Vásárhelyi—the method used today in most air traffic control systems, where a group of human controllers oversees flight patterns.
The experiment used two layers for the traffic, spaced approximately 46 feet apart. Travel speed was set to around 20 feet per second horizontally and 5 feet per second vertically. The researchers tuned their system to avoid “ghost jams,” which occur when individuals in the system seem to slow for no reason—thereby creating a ripple that further disrupts the traffic. Such jams are an emergent property of decentralized traffic systems, and a lot of work has gone into determining ways to avoid them.
Aeronautical engineering professor Mirko Kovac, the director of the Aerial Robotics Laboratory at Imperial College London, notes that efforts to harness the “swarm intelligence” of large groups of drones is one of the next frontiers in the technology, with growing importance in many different fields, including logistics, construction, and environmental sensing.
Kovac, who wasn’t involved in the new study, says such drone traffic management systems will be “an important part of a future where ecosystems of robots work together and integrate in our man-made environments.”
As commercial drones take to the skies in ever-increasing numbers, avoiding aerial gridlock and in-air collisions among them will become a significant issue. It makes sense that for solutions, we should look to the birds, which have been flying in flocks for millions of years.
Lead image: Andy Dean Photography / Shutterstock
Get the Nautilus newsletter
Cutting-edge science, unraveled by the very brightest living thinkers.
