This Smart Traffic Management System Can Handle 5,000 Autonomous Drones, Researchers Say
With the number of drones deployed ever rising, new approaches to handling the traffic are required — and this team says it has the answer.
A research team at the Eötvös Loránd University (ELTE), working with CollMot Robotics, has developed a way to better handle large volumes of autonomous drones safely — proving its efficacy with 100 drones in the real world and up to 5,000 in simulations.
"Coordination of local and global aerial traffic has become a legal and technological bottleneck as the number of unmanned vehicles in the common airspace continues to grow," the team explains of the problem it set out to solve. "To meet this challenge, automation and decentralization of control is an unavoidable requirement."
"In the US," the team notes, "the FAA [Federal Aviation Administration] currently registers around 200,000 manned aircraft generating only a couple of thousand flights simultaneously at peak times, while there are already over half million recreational and over 300,000 commercial drones registered with over 240,000 pilots who wish to use the same airspace as flexibly and as frequently as possible."
Building on earlier work in collaborative swarm control, the research team set about creating a traffic control system that would scale up to the densities of drones predicted to be seen in future smart cities — effectively an automated air traffic control system. Where previous efforts have concentrated on scenarios where the drones are working on concert towards a common goal, the new system addresses the potential for drones to be operating independently.
Using a forward-thinking route planner combined with flocking models inspired by nature, the team's traffic control system is designed to allow the drones to avoid conflict even when operating independently — either by routing them around other drones in advance or coordinating with nearby neighbors if a collision is imminent. The system was tested in simulations of up to 5,000 individual drones across two-dimensional airspace, with the potential to layer the system for three-dimensional traffic control; a real-world demonstration, meanwhile, showed it working with 100 physical drones.
"Aerial missions with more than a single drone involved require coordination between the cooperating agents, which can be solved now with our algorithm," the researchers claim. "An extended version of our collision avoidance strategy can also be part of a global UTM [Unmanned Aircraft Systems Traffic Management] solution if autonomous agents get equipped with the proper communication devices and onboard DAA [Detect And Avoid] control units running the algorithm.
"But," the researchers continue, "due to its low computational cost, the decentralized traffic model can also be used in a centralized way in air traffic management as an automated planner or recommendation system for human air traffic controllers."
The team's work has been published under open-access terms in the journal Swarm Intelligence.