Engineers and researchers from the Georgia Tech Research Institute and Naval Postgraduate School in the United States of America have pitted two swarms of autonomous aircraft against one another in a test of dogfighting for drones.
The exercise over a military test facility in southern California is the first example of a live engagement between two swarms of unmanned air vehicles (UAVs), allowing the two teams to demonstrate different combat tactics in flight.
According to Don Davis, division chief of the Robotic and Autonomous Systems Branch of the Georgia Tech Research Institute, defense players are interested in the ability to engineer autonomous swarms of UAVs that can engage another swarm of hostile UAVs.
"This experiment demonstrated the advances made in collaborative autonomy and the ability of a team of unmanned vehicles to execute complex missions," he said.
"This encounter will serve to advance and inform future efforts in developing autonomous vehicle capabilities."
While aerial dogfighting tactics have advanced dramatically since World War I, its application to UAV swarms may bring about a new set of challenges, because unmanned vehicles can dive, bank, and climb at rates human pilots cannot tolerate. It also brings to powerful computing capabilities to bear on the challenge, with the ability to track dozens of adversaries.
The researchers are leveraging machine learning to optimise performance and help the unmanned aircraft recognise under which circumstances a particular tactic may be advantageous. They are also looking into ways individual aircraft in a swarm can autonomously coordinate themselves and engage in tactical behaviour.
The trial sorties, which took place at Camp Roberts, a California National Guard facility, happened after some refinements of the tactics. First, the teams rapidly tested tactics on a simulator that runs 30 times faster than real time. They then took promising approaches and tested them on a full software stack that includes a high-resolution simulation.
The full software stack includes the autonomy logic, communications systems, collaboration algorithms and other software that is then inserted directly into the actual aircraft.
Each team launched ten small propeller-driven Zephyr aircraft, though two of the aircraft experienced technical issues at launch and were unable to compete, resulting in a 10 versus 8 competition. Although the UAVs were physically identical, their computers used different autonomy logic, collaboration approaches, and communications software developed by the two institutions.
The aircraft were privy to the location of the others thanks to GPS tracking, but in the future, this information will be provided by on-board cameras, radars and other sensors and payloads.
On each aircraft was a single-board mission computer, with flight controlled provided by an open-source autopilot. The aircraft also had WiFi systems, so they could communicate with other aircraft and with a ground station.
While each aircraft was fully controlled by the autonomy algorithms, a safety pilot was on standby to take control of any aircraft if necessary. The autopilots also had built-in safety constraints, such as airspace boundaries and ranges.
According to the researchers, while both teams were trying to solve the same problem, they came up with solutions that were similar in some ways and different in others. By comparing how well each approach worked in the air, they could refine strategies and tactics.
While the drones were not able to actually shoot at each other, a ground computer determined when they were in a position where they could attack another aircraft.
The trial was a success.
"It gave us, as far as I know, the first actual experimentation of flying two autonomous swarms of UAVs against one another with no human control, other than sending high level commands or sending a message to engage," explained Davis.
"We were really trying to understand how different autonomy tactics work against other autonomy tactics."
[Nominations are now open for the Engineers Australia Sir George Julius Award, AGM Michell Award and Young Mechanical Engineer of the Year Award. Find out more.]