RMIT drone can think like an airline pilot

RMIT drone can think like an airline pilot

Drone equipped with technology that allows it to communicate with an air traffic controller

Researchers at RMIT University have developed a drone equipped with technology that allows it to automatically communicate through human language with an air traffic controller if it loses connection with a remote pilot on the ground.

“The air traffic controller can issue clearances or tell it to hold or anything required for it to travel safely in the airspace. And it can do that exactly like a main pilot would, so it has the on-board decision making and other aspects that allows it to think and act like a pilot,” said Dr Reece Clothier, leader of the RMIT Unmanned Aircraft Systems (UAS) Research Team.

“We have provided the drone the ability to talk and think itself should the remote pilot on the ground not be able to talk to an air traffic controller. And that’s quite a common situation. You have communication outages between the ground station and the drone, for whatever reason the communication can be lost with the drone itself.”

The drone or unmanned aircraft is able to communicate to an air traffic controller through English. It’s based on UFA’s ATVoice technology, which uses the International Civil Aviation Organization (ICAO) standard of phraseology, but the RMIT researchers are looking into more advanced machine learning algorithms for it handle more complex decision making.

“The drone can interpret messages that are issued to it, and decide whether it can actually respond to a clearance issued to it correctly … and decide whether it can safely act on those clearances.

"We’ve had our own staff mimic the role of the air traffic controller and phone up the drone and say ‘I would like you to hold here’ and our drone has responded and executed those clearances as requested.”

At the moment the drone is not highly intelligent, Clothier said. It has fairy static checks and responses and can only carry out basic decision making.

“A pilot would have to make decisions as to the safety of his or her flight, and that may mean ignoring air traffic control advice or asking for corrections or changing requests because there might be sudden changes in the environment and those sorts of thing.

“That requires a high order of artificial intelligence and decision making. So there’s plenty of avenues for further research there to improve robustness of the system, to be more and more like a human pilot,” he said.

Clothier said the drone is not designed to replace the human pilot as it would kick in or assist when communications on ground level are lost. He said he sees it being used as a supplementary piece of technology to the pilot.

Clothier and his team of researchers are also looking into the psychology of air traffic controller and pilots interacting with the technology as part of their research.

“We don’t really know how air traffic controllers will respond to an automated machine talking back to them. So it’s not just about the engineering and design, it’s also about how the human interacts with the device.

“So we get to explore those, and whether the system reduces the task of the workload placed on the remote pilot and likewise the workload factors associated with the air traffic controller. We’ve only just developed the technology but now we need to say if we implemented this and had if flying all over the world, what would be the human issues around it?”

The team presented their research paper at the Australian International Aerospace Congress this week, which will be made available in the coming weeks, Clothier said.

The prototype was tested late last year with Thales’ Top Sky Air Traffic Control System.

Join the CIO Australia group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.

Join the CIO newsletter!

Error: Please check your email address.

Tags dronemachine learningvoice recognition

More about RMITRMIT University

Show Comments