UK researchers start using AI for air traffic control

Receive free AI updates

British researchers have created a computer model of air traffic control in which all flight movements are directed by artificial intelligence rather than humans.

Their “digital twin” of the airspace over England represents the initial outcome of a £15m project to identify the role artificial intelligence could play in providing advice and eventually replacing human air traffic controllers.

The research, known as Project Bluebird, is a partnership between the National Air Traffic Service, which is responsible for UK air traffic control, the Alan Turing Institute, the national data science and artificial intelligence agency, and the University of Exeter, and received government funding from UK Research and Innovation, a national agency. Its first results were presented at the British Science Festival in Exeter.

Reasons for incorporating AI into air traffic control include guiding aircraft along more fuel-efficient routes to reduce aviation’s environmental impact, as well as reducing delays and congestion, particularly at busy airports such as London Heathrow.

There is also a shortage of air traffic controllers, which take three years to train.

Nats has a more complete database of past flight records than any other air traffic control agency in the world, which researchers are using to train their artificial intelligence systems, said Richard Everson, a professor of machine learning at the University of Exeter.

“We have been preparing for this for the past ten years, recording air traffic activity over the UK,” said Richard Cannon, head of bluebird research at the National Team. The data includes 10 million flight paths.

Human controllers and AI agents are now working together to handle aircraft in the project’s digital twin of UK airspace, using accurate simulations of real-life air traffic.

“By the end of the project in 2026, we aim to conduct live ‘shadow trials’ in which AI agents will be tested in real time against air traffic data to directly compare with the decisions of human air traffic controllers,” Cannon said. .

But he emphasized that the artificial intelligence system does not have the authority to actually determine the aircraft’s route.

If the research is successful, it could first lead to wider operational trials of artificial intelligence with people within a few years, before national air transport agencies and other air traffic agencies consider introducing computer-controlled systems.

Nats already has powerful computer systems that can process data in one of the world’s most congested airspaces, but has not yet used artificial intelligence to predict future flight trajectories.

The system failed over last month’s bank holiday weekend when it failed to recognize flight plans with conflicting data, causing travel chaos. The Civil Aviation Administration of China has launched an investigation into the cause of the problem, and the airline has requested compensation from the National Air Force.

Cannon and Everson said they could not comment on whether artificial intelligence could have prevented a system failure that resulted in the cancellation of more than 1,500 flights.

But Everson said AI should make air traffic control more resilient and reduce the risk of it failing in the face of unexpected events.

The digital twin built by Bluebird covers the so-called London Flight Information Area, which covers the airspace across much of England and Wales.

Iverson said the artificial intelligence system is routing the aircraft with the necessary separation of 1,000 feet vertically and 5 nautical miles horizontally.

It also ensures that each flight path remains safe from collision for at least 15 minutes if radio communications with pilots fail for any reason.

“In every air traffic control system in the world, all decisions are made by humans,” Cannon said. “We’re not saying we want to automate the skies over the UK, but we are pushing the envelope as safely as possible.”

Svlook