AI and data analytics discussed at aviation training conference

By Shane Klestinski, Associate Editor

 

Subject matter experts shared their insights in a session focused on artificial intelligence (AI) and data analytics during the World Aviation Training Summit (WATS) in Orlando, Florida, April 8.

 

WATS brings together leaders from throughout the airline training community. Its organizer, Halldale Group, describes WATS as “the world’s largest gathering of aviation training professionals, serving airlines, aircraft manufacturers, regulators, training providers and the training industry, designed by and for aviation professionals.”

 

Participants who took turns presenting their findings included: Stéphan Labrucherie, head of flight training worldwide at Airbus; Thomas Bessière, CEO of HINFACT; Brock Booher, president of Volant Systems; Shane Jordan, Turtle Creek LLC president/lead developer; and Neil Waterman, commercial aviation director with ASTi.

 

Labrucherie and Bessière began the session with their presentation, “Tracking, Data & Us,” discussing competency-based training and assessment, as well as pilot competence and resilience. Their talk involved various levels of grading metrics that indicated a trainee’s progress, and, Labrucherie noted, “how we will be able to tailor and customize training based on evidence,” depending on the context of training.

 

“We have some solutions to make what we call ‘profiling’ based on all the data we can collect from the training and simulators to identify the strengths and weaknesses of each individual or group of individuals at the operator level,” Bessière said. “Here, we can [identify shortcomings] and suggest the right scenarios at the right moment, so we move from static training to something more dynamic and adaptive. If we extend this profiling with training data and all the data, we collect from operations… it’s also a way to fit the system with a new source of data to better assess the strengths and weaknesses of the pilots.”

 

Booher and Jordan’s talk, “AI = Aviation Intelligence: Shaping the Future of Aviation Training,” described different aspects of AI applications to aviation training. Booher began by describing the creation of the first operational radar site in 1937 – which, according to Booher, actually started with a contract to build a “death ray.” Today, radar use is worldwide, employed in air traffic control, weather detection, target acquisition, and several other areas. Booher compared this brief history of radar to the recent technological shift with AI.

 

To demonstrate this point, Booher showed a video of an avatar based on himself, delivering the next few lines of his presentation using various foreign languages – which he doesn’t speak – but AI made possible in under an hour.

 

“Imagine the possibilities for your training organization producing materials in various languages,” Booher said. “AI has the potential to accelerate learning by personalizing the experience for each learner and adapting and responding to their individual needs.”

 

He went on to impress some points on what is possible, permissible and (financially) practical with AI, and Jordan’s portion elaborated on some of the practical aspects.

 

“As with most large infrastructure projects, you need to start by defining what is ‘success’ and what will that look like?” Jordan said. “That will help avoid pitfalls and hidden costs. One thing we’ve learned is that up-front development costs can be quite expensive, especially if you’re exploring. It’s better to iterate quickly and fail quickly to learn what will work.”

 

Jordan advised the audience that data management “is one of the more expensive sides of AI.” The required data is large and depending on what is required for a specific use case, users should consider how the AI will access storage. He also recommended that users consider the transactional cost between the AI and their service.

 

In “AI: Powering-up Pilot Training – A Case Study of SATCE and Virtual Instruction,” Waterman discussed AI in aviation training environments with a focus on the Simulated Air Traffic Control Environment (SATCE). He also listed some problems in pilot training that other Team Orlando members have publicly mentioned in the past, such as a shortage of trained pilots, unmet training throughput goals, and constrained budgets. Waterman revealed that, in a particular case involving the Navy, virtual reality-based trainers as part of a self-paced training program provided a solution.

 

“Students could sign up for themselves, they really liked the idea,” Waterman said. “They just booked a device and picked the next lesson that was selected through the underlying learning management system, and they’d just go and use it. They didn’t need an instructor to be there, if a device was available, they’d use [one of 50]. In fact, the Navy was so happy with the SATCE part of this [program] that they’ve added a number of other devices since then.”

 

Waterman finished his description of this use case by saying that the Navy’s pilot graduation rate went from 91% to 101% in the first year, due to “recycled” students who had successfully graduated since beginning the new approach. He added that the current 50 devices were getting increased to 150.

 

“They’ve seen the effectiveness and they’re growing the use of it,” Waterman said. “It shows the power of technology. [The Navy] likes it because it works. The students like it because it makes them more proficient, and they feel better when they move from the simulator up to the flightline because it doesn’t feel like much of a transition.”