“Lexi, Book Me a Jet”: How AI in Aviation is Evolving

AI in Aviation

At EBACE 2019, industry experts examined the growth of artificial intelligence (AI) in aviation and how it will evolve in the future. In nearly every industry—including aviation—AI and machine learning are having an impact. In fact, the data collected within the connected aviation ecosystem is opening the door to many new and useful applications.

In a panel at EBACE, Josh Gelinske of Appareo, Kurt Doughty of Collins Aerospace, Peter Conrardy of GE Aviation, and Bernhard Fragner of Global Air joined moderator Eric Leopold from the International Air Transport Association (IATA) to talk about AI in aviation. The discussion focused on how deep learning and AI can be used in business aviation and the effects it will have on the industry as a whole.

AI in Aviation
Kurt Doughty of Collins Aerospace

“Machine learning and AI technology has been around for a long time, yet as other technologies evolve, machine learning and AI are becoming more applicable to the aviation industry,” Doughty said in a recent interview with us. “As computational power, sensing, and data transmission capabilities increase, the size and types of the data sets we can capture and analyze become larger, which enables the use of machine learning and AI to transform the data into actionable intelligence.”

The aviation industry is large and complex, creating numerous opportunities where AI can make a significant difference. Use cases could include ground traffic optimization, flight planning, crew planning, and predictive maintenance for the aircraft to avoid operational disruption.

“One of the biggest costs to an airline is an unexpected event that causes a disruption,” Doughty said. In his role at Collins Aerospace, Doughty is focused on enabling a more efficient, intelligent aircraft. “By aggregating data from various sources such as full flight parametric and fault data from aircraft operators and combining it with third-party data sets for things like weather or pollution, as well as our own internal data sources, we can enable more intelligent operations,” he notes. “As processing power increases, these massive amounts of data can be quickly analyzed to support near real-time decisions. But the key factor to successful implementation of AI is asking the right questions and understanding the business outcomes operators hope to achieve.”

There are, however, a few challenges that the industry needs to address before getting to the point of asking, “Lexi, book me a jet.”

Confidence in Adoption:

The hesitation for adoption isn’t unique and one that has affected the industry as new elements of flight have been impacted by autonomous systems. Yet, in each of these cases, like with autonomous steering, the systems were developed in a more controlled setting. “The acceptance and adoption of AI will eventually come as we see more precise, higher quality, and robust outcomes that AI will produce, and the true value of the data and AI applications is demonstrated,” Doughty told us.

An Element of Control:

When considering using machine learning and AI to address a challenge, it is important to have the controls in place and an understanding of the boundaries that are created for which the AI can access data. Doughty explained that there is a need to ensure that the AI plays within the boundaries that are set and doesn’t go beyond the parameters to find and process data.

Ensuring Cybersecurity:

As with any system that maintains data sets, it is critical to ensure that the data is secure. Cybersecurity threats could impact the integrity of the data, by injecting malicious data into the data set and thereby impacting the intelligence that is returned. Proper security controls at the highest level must be in place. “The system is only as good as the input that it receives,” Doughty notes.

As these challenges are being addressed within the industry, there continues to be great advancements in computational processing of large data sets, aggregation of many data sources from third-parties, and implementing AI for use cases that will ultimately create a more intelligent aviation ecosystem.