The research is part of the Tech’s research institute’s (GTRI’s) efforts to reducing the load on human remote operators by improving getting autonomous systems to work in teams.
“For autonomous systems to scale effectively, future systems will need the ability to perform with a higher level of autonomy,” said GTRI chief scientist Lora Weiss (pictured with the aeroplanes). “Human operators must be able to provide high-level task descriptions, allowing the systems to figure out for themselves how to dynamically form teams and autonomously collaborate to complete tasks.”
A single UAV was initially designated as the leader and commanded to fly autonomous figure-eight patterns.
Two follower planes joined the first at the same altitude, holding relative position, separated by around 50m.
The leader shared its current position with the followers several times per second allowing the followers to update their own autopilots and adjusted their controls to reach the desired position.
GTRI’s algorithms are general enough to be used with different UAVs and autopilot systems.
In the trial, quarter-scale Piper Cubs were used (wingspan ~ 2.4m) carrying a mission computer, autopilot, and sensors.
Capabilities, said Weiss, could include automatically splitting up a search area between many
Applications, said Geargia Tech, could include providing several different camera angles while imaging, or sharing different sensors for a particular task.
The recent tests follow a 2010 GTRI demonstration that featured two small-scale aircraft and a full-size self-driving car.
There is a video of Georgia Tech’s collaborating UAVs.