For decades, in the world of professional baseball in America, the selection process to identify players most likely to be successful in the “major leagues” began with a small group of colorful characters known as “scouts.” Usually older, perhaps former players at some point in their lives, they are portrayed in movies as tobacco-spitting, sharp-eyed judges of prospective talent. They would roam the dusty roads of small towns across the USA in the quest to discover the unknown phenom who could throw the ball 100 miles per hour or hit it 500 feet. It was a pure form of human factors analysis.
Today, as a profit-driven entertainment business, baseball clubs (and many other sports) are turning more to “scorers” who rely on “data capture” of a player’s myriad and often esoteric statistics. Past performance is used to, hopefully, predict future behaviour. (And how big a signing bonus to offer.)
Airline training is undergoing a similar transition. Data analytics are moving to the fore, from the selection process for cadet candidates to type-rating and recurrent training for line pilots.
In the transition, the role of instructors (the scouts) is changing. Instead of monitoring how far off the glide slope the crew has strayed in a flight simulation session, the data is captured and dissected by software (the scorers). Objective parameters such as excessive control inputs, point of touchdown, and so forth can be assessed against a benchmark; airlines can identify which measurements are priority to their standards. Trends across the pilot community can be identified, and the training curricula adjusted to address frequent out-of-range issues.
Some airline training organisations are also employing eye-tracking technology to determine where a pilot is looking (and for how long) at critical moments during emergency scenarios.
At the moment, most of the attention on data analysis is on simulator statistics. But the technology exists to capture similar information from operational flights and compare it with training session data. Assuming, of course, the pilot unions assent to the proposed privacy provisions.
Researchers are also exploring applications of machine learning/artificial intelligence to predict what will happen, based on analysis of the simulator/aircraft attitude, etc., and the pilot’s data-identified tendencies.
So what of the instructor now? Has she or he been supplanted, their years of experience abandoned, retired to the bleachers like the old baseball scout?
Not at all. The instructor can now focus more attention on so-called “soft skills” such as communication between the Captain and the FO. Moreover, the data capture will provide information which is not even visible to the instructor in the jump seat. They will no longer be tethered to the IOS monitors and mental math gymnastics, but can instead engage with the crew as needed and truly draw on their depth of flying experiences.
In effect, instructors, who have historically been both scouts and scorers, can now concentrate on their strengths of observing and assessing pilot behaviour and crew coordination.
Whether scout- or scorer-driven, baseball’s track record is not stellar. Of the thousand or so players drafted each year, fewer than 10% ever play in the majors. Airline training is somewhat better – airline heads of training estimate 50% of training school graduates are sufficiently qualified to begin flying passengers – but that ratio needs to get much better if the training industry is to keep up with the predicted demand for pilots over the next couple of decades. Perhaps the emerging approaches to data analysis, as they are refined, together with the retained expertise of instructors, will improve the quality and quantity of pilots.
Rick Adams, CAT Editor
Published in CAT issue 4/2019