Pilot shortage ? Hal9000 will take care ...
“ Airbus Accelerates AI Ahead Of Looming Pilot Shortage
Guy Norris
As the civil aerospace industry looks increasingly to artificial intelligence (AI) to help improve safety standards, and amid the looming potential for a global pilot shortage, Airbus is stepping up research into autonomous systems and the potential for single-pilot operations.
As part of these efforts, the manufacturer has created Airbus AIGym, a platform that is being used to evaluate the use of AI and machine learning to tackle real issues across the enterprise—including autonomous systems for the flight deck. Through the initiative, Airbus is encouraging research in specific areas such as speech recognition by issuing a series of challenges that are open to companies, research laboratories, academia and individuals.
The air traffic control (ATC) speech-recognition challenge, which closed late in 2018, will be followed shortly by another aimed at devising AI systems capable of identifying and recognizing taxiway signs. The work supports the development of autonomous taxiing capability and is designed to feed into the company’s Automated Takeoff and Landing (ATTOL) demonstration program, which is planned for flight tests in the mid-2020s. The project is intended to leverage computer vision technologies and techniques to enable an aircraft to navigate and detect obstacles during taxi, takeoff, approach and landing.
“How do we reduce the workload on pilots as airspace becomes more congested? How do we continuously improve safety and cope with the envisioned shortage of pilots?” asks Airbus Chief Technology Officer Grazia Vittadini. “Today we have 200,000 type-rated commercial pilots. If we apply market projections, we will have around 50,000 aircraft in service after 2037. That means we will need 600,000 pilots over the next 20 years.”
The threat of an industry-wide pilot shortage is “pushing us toward automated systems with the alternate ambition to go for autonomy powered by artificial intelligence,” says Vittadini. “That’s a challenge not to be underestimated, especially for certification. The certification of systems based on nondeterministic inputs is not trivial. It is a page of aviation we have not yet written, but we are starting to write it now with partners and regulators,” she explains.
Airbus believes areas such as communications and navigation represent early potential opportunities for AI to support or supplement current cockpit tasks. The recently concluded contest to develop ATC speech recognition, for example, focused on areas such as standard audio transcription and call-sign extraction and produced “tangible results” from 25 finalists, says Vittadini. “But it is just the beginning, and we are about to launch another challenge along those lines to identify and recognize taxiway signs,” she notes.
In a related area, Airbus Defense and Space already is working with mobility provider ZF Friedrichshafen to enhance an autonomous system for self-driving cars and self-positioning vehicles using ground-control points calculated using Airbus-derived satellite data. For the work in Germany, the positioning information is being combined with lidar and radar data provided by ZF sensors.
Autonomous sense-and-avoid system technology for ATTOL is being developed by Wayfinder, a project spun out of the Silicon Valley-based Airbus A3 team working on the Vahana electric vertical-takeoff-and-landing demonstrator. “Wayfinder is developing machine learning and one of the bricks of the more automated and autonomous systems work for ATTOL,” says Vittadini.
“We plan to solve the most challenging problems in developing scalable, certifiable autonomy systems to power self-piloted aircraft applications throughout Airbus, from small urban air taxis like Vahana to large commercial airplanes,” Wayfinder's director of engineering, Alex Naiman, says in a recent blog post.
Describing the individual challenge areas as technology “bricks” that will provide the foundation for future demonstrations and development, Vittadini says that beyond the ATTOL program lies the company’s longer-term ambition to create a reduced-crew aircraft and, ultimately, a fully autonomous aircraft. But she cautions that it will take a long time to reach these targets.
“Let’s not forget it took us 60 years to go from four-crew cockpits to two-crew, so we won’t get there anytime soon,” Vittadini notes. “But yes, it is an ambition and we have planned the steps. So we will go from ATTOL to reduced-crew operations in which we can have single-pilot operations in certain phases of flight such as during cruise.”
The initiative is supported by work on a “Disruptive Cockpit” simulator at the company’s Toulouse headquarters. Developed as part of Europe’s Clean Sky 2 research program, the project is focused on demonstrating new functions, computing resources and communications. These include cockpit procedure automation, pilot-monitoring systems such as eye-tracking, head-worn displays, ground collision avoidance, new navigation sensors, voice-recognition systems for communications with ATC and airline operations, multimodal integration for flight crew interfaces, tactile head-mounted instruments and image-based landing systems.
“The flight deck is accessible; we are testing it with our own test pilots and customers. It’s a journey and a long journey, but safety is first, and we will take it step by step,” Vittadini adds. “We are already testing different bricks of a possible single-pilot operation in our enhanced multicrew cockpit where we study what it would mean in cruise to have the copilot resting in his or her seat and an active captain in the left seat. So we are studying several potential combinations of solutions “