“The ability to engage a swarm of threat UAVs with another autonomous swarm is an area of critical research for defense applications,” Don Davis, division chief of the robotics and autonomous systems branch of the Georgia Tech Research Institute, said in a press release. “This experiment demonstrated the advances made in collaborative autonomy and the ability of a team of unmanned vehicles to execute complex missions. This encounter will serve to advance and inform future efforts in developing autonomous vehicle capabilities.”
The teams each attempted to launch 10 drones, but two drones failed to take off, so the dogfight was uneven. Though the drones were identical in form, the algorithms that controlled the aircraft differed, allowing them to fly in varying formations and test a number of tactics.
“Both teams were trying to solve the same problem of flying a large swarm in a meaningful mission, and we came up with solutions that were similar in some ways and different in others,” said Charles Pippin, a senior research scientist at the Georgia Tech Research Institute. “By comparing how well each approach worked in the air, we were able to compare strategies and tactics on platforms capable of the same flight dynamics.”
Before taking the drones out for a live engagement, the researchers tested their algorithms in a series of rapid simulations that helped determine the most effective tactics. The researchers hope their tests will help identify how aerial battles might best be fought in the future.
“Autonomous techniques using machine learning may identify new tactics that a human would never think of,” Davis said. “Humans tend to base their techniques on tactics that manned fighters have used in the past. These autonomous aircraft may invoke new strategies.”
- The best free flight simulators
- Want to control a swarm of drones? There’s an app for that
- The best Cyber Week drone deals for 2020
- The best Black Friday drone deals for 2020
- How A.I. bumblebee brains could usher in a new era for navigation