Like a trainee pilot, with the best instructors.
Nobody wrote rules telling it how to fly. It learned by example, in four steps.
-
A pilot flies
Experienced pilots fly the simulator while the computer watches every move they make.
-
It takes notes
It records what the pilot did, and what was happening at that moment: speed, height and wind.
-
It practises
It learns which action fits each moment, until it can choose the right one by itself.
-
It flies on its own
Then it takes the controls, from take-off to landing, even in emergencies and bad weather.
Research you can rely on.
Safer when it matters most
An autopilot that copes with emergencies and severe weather could support pilots in the moments that are hardest for people.
Science anyone can check
Our results are published in five peer-reviewed papers, so independent experts can examine them.
Behind everything we build
The same research-first approach shapes all our work, from cameras that spot hazards to AI that helps doctors.
Five milestones, one flight.
Each milestone is a peer-reviewed research paper, from the first lessons to complete flights.
-
LEARNING FROM PILOTS
It learned to fly by copying experienced pilots.
The first step: basic flying skills, learned by watching real pilots at the controls.
THE PAPER · IEEE An Intelligent Autopilot System that learns piloting skills from human pilots by imitation Read the paper ↗ -
HANDLING EMERGENCIES
It learned what to do when something goes wrong.
Pilots train hard for emergencies. The autopilot learned how to handle them in the same way: from pilots.
THE PAPER · IEEE An Intelligent Autopilot System that learns flight emergency procedures by imitating human pilots Read the paper ↗ -
FINDING ITS WAY AND LANDING
It found its way and landed an airliner by itself.
The autopilot navigated an airliner to its destination and brought it down to land, with no one at the controls.
THE PAPER · IEEE Autonomous navigation and landing of airliners using artificial neural networks and learning by imitation Read the paper ↗ -
LANDING IN BAD WEATHER
It landed in severe weather, and knew when to try again.
In severe weather, it landed safely, and knew when to abort a landing and go around for another attempt.
THE PAPER · IEEE Autonomous landing and go-around of airliners under severe weather conditions using artificial neural networks Read the paper ↗ -
BEYOND TODAY’S AUTOPILOTS
It handled flights that today’s autopilots can’t.
Complete flights from take-off to landing, including extreme landings beyond the limits of the autopilots in service today.
THE PAPER · SPRINGER · APPLIED INTELLIGENCE Autonomous flight cycles and extreme landings of airliners beyond the current limits and capabilities using artificial neural networks Read the paper ↗