This article is published in Aviation Week & Space Technology and is free to read until Aug 28, 2024. If you want to read more articles from this publication, please click the link to subscribe.
An Embry-Riddle student practices radio communications by interacting with AI-driven air traffic control and pilots in the university’s Preflight Immersion Lab.
As artificial intelligence technology takes off, it is natural to ask how it might be employed in pilot training.
One area where AI can enhance current practices is in learning pilot-to-pilot and pilot-to-air traffic control communications—everyday activities critical to aviation safety.
Teaching pilot communication skills has followed a timeworn path: an abundance of repetition and practice in the cockpit or in simulators. This poses a challenge for instructors, who are unable to stop and review each communication to provide corrective feedback midflight. Students also face the stress of live training flights when they must suddenly navigate real-time, rapid-fire radio communications among multiple parties.
Embry-Riddle Aeronautical University is immersed in comprehensive methods for using AI to teach students standard phraseology before they start to fly, without instructor role play and in a way that provides active assessment and better clarity.
AI’s automatic speech recognition, speech-to-text generation and natural language processing capacities allow for the interpretation of human speech and the response to it, as well as the ability to give corrective feedback—all in real time. “Without that, all you have is this very passive environment for independent practice,” Aviation English Coordinator Andrew Schneider says.
Two systems provided by our industry partner Advanced Simulation Technology Inc. (ASTi) form the basis of AI-driven air traffic control (ATC) communications training at Embry-Riddle. The ASTi systems have been tailored to mimic the aircraft, airspace and flight scenarios that students encounter, including the busy traffic at Florida’s Daytona Beach International Airport.
Students begin in a virtual reality (VR) setup, receiving radio communication lessons from an instructor during a 360-deg. flight video. They then review the communications with ASTi’s Pilot Phraseology Trainer, a platform that narrows the focus of radio calls.
The trainer shows students images of instruments, flight diagrams and 3D models. Using headsets, students listen to ATC and respond to the various prompts and visuals. AI technology interprets their responses and highlights errors. Each response is also scored based on pronunciation and structure, as well as adherence to FAA communication standards. Students practice the radio calls step by step and at their own pace by interacting with the AI.
Students then proceed to a system that fuses a customized VR flight trainer with ASTi’s Simulated Environment for Realistic ATC (SERA). SERA uses machine learning to create a dynamic environment filled with simulated air traffic controllers and radio communications from other aircraft. The controllers can give instructions—on position, altitude and route, for example—and respond to student requests, querying them when communications are not understood, just as real-life controllers would.
First-year aeronautical science students spend 9 hr. on radio communications as part of a front-loaded, four-week immersive flight-training program, which also includes learning preflight, checklist and flight procedures in VR environments.
Ken Byrnes, chair of the flight department at the Daytona Beach campus, says the time it takes for students to become proficient in radio communications has been substantially reduced since the program started. There is “significant improvement in their communications performance from their first flight because we’ve spent so many hours focusing on it in a nonthreatening environment,” he adds.
Schneider, who designed the ATC course, is developing the next step in ATC education, employing speech recognition and natural language processing tools on live flights. The leap forward will be to go beyond communications training to apply AI technology to the massive, complex data sets collected at Embry-Riddle, which counts some 1,400 active flight students at any time.
An AI tool trained on data from novice and expert pilots could help detect flight-skill mastery, notes Tom Schnell, director of the University of Iowa’s Operator Performance Laboratory. Using performance data from simulated or live flights, this tool could create scenarios and lessons that reinforce training in areas where the student is weaker, acting as a “companion system” for the flight instructor, Schnell says.
During tours of Embry-Riddle’s immersion lab, visitors are often impressed by students in headsets running through checklists, moving flaps and flying. What is just as astounding, though, is to hear them—and to realize that these students are communicating throughout an entire flight without ever having stepped into one of our aircraft.
P. Barry Butler is president of Embry-Riddle Aeronautical University.
Related Content
