Airbus Starts Testing Autonomous Landing, Taxi Assistance on A350 DragonFly Demonstrator
Airbus UpNext, a wholly owned subsidiary of Airbus, has started testing new, on ground and in-flight, pilot assistance technologies on an A350-1000 test aircraft.
Known as DragonFly, the technologies being demonstrated include automated emergency diversion in cruise, automatic landing and taxi assistance and are aimed at evaluating the feasibility and pertinence of further exploring autonomous flight systems in support of safer and more efficient operations.
“These tests are one of several steps in the methodical research of technologies to further enhance operations and improve safety,” said Isabelle Lacaze, Head of DragonFly demonstrator, Airbus UpNext. “Inspired by biomimicry, the systems being tested have been designed to identify features in the landscape that enable an aircraft to “see” and safely maneuver autonomously within its surroundings, in the same way that dragonflies are known to have the ability to recognize landmarks.”
During the flight test campaign, the technologies were able to assist pilots in-flight, managing a simulated incapacitated crew member event, and during landing and taxiing operations. Considering external factors such as flight zones, terrain and weather conditions, the aircraft was able to generate a new flight trajectory plan and communicate with both Air Traffic Control (ATC) and the airline Operations Control Centre.
Airbus UpNext has also explored features for taxi assistance, which were tested in real-time conditions at Toulouse-Blagnac Airport. The technology provides the crew with audio alerts in reaction to obstacles, assisted speed control, and guidance to the runway using a dedicated airport map.
In a related post, Acubed, the Silicon Valley-based research arm of Airbus, shared some details on how its engineers are contributing to the DragonFly Demonstrator initiative. As a key partner and contributor to the DragonFly demonstrator, our Acubed team is tackling the following challenges:
- Autonomy functions require an entirely new, data-driven development approach. These functions operate in a complex and changing environment. The challenge comes when attempting to capture this complexity and building a reliable AI algorithm.
- Seeing is believing when it comes to unlikely events. To create a robust, certifiable algorithm, Airbus will have to capture enough “corner case” data to deliver an algorithm that can perceive and react to an extremely unlikely event, even as unlikely as an elephant on the runway - a case found on an African airport.
- Data. Data. How to collect at scale? Developing AI to robustly understand the environment and guide the aircraft during landing and taxiing will require massive amounts of data. Acubed is using its own aircraft, a Beechcraft Baron 58, to fast track the development of its data collection system and data management infrastructure, targeting scales in the tens of Petabytes.
Several external industry partners including Cobham, Collins Aerospace, Honeywell, Onera and Thales are also supplying technologies for DragonFly flight testing. DragonFly was partially funded by the French Civil Aviation Authority (DGAC) as part of the French Stimulus plan, which is part of the European Plan, Next Generation EU, and the France 2030 plan.