Will the AI Revolution Include Business Aviation?

Many proponents of the rapidly developing software technology known as artificial intelligence have already christened AI as a kind of magic bullet that promises to improve just about every segment of our lives. The precise roles this emerging technology might ultimately play in the business aviation sector remain largely unclear, but experts point to enhanced drone operations, optimised routes, supercharged documentation and maintenance as some of the initial areas where AI might be useful.

First, however, it may be helpful to quickly discuss what AI is and what it is not. At its most basic level, AI is a term that broadly covers rapid advancements in computer processing performance. It enables systems to crunch larger amounts of data faster and compare statistical information in real-time. Data processors can do it all so fast that they can virtually mirror human abilities such as pattern recognition, basic reasoning and problem-solving.

“In developing our autonomous piloting systems, Daedalean deploys a subfield of AI called machine learning,” explained the Switzerland-based company’s Director of Communications John Mora. “We use it for interpreting visual information, which creates capabilities for our system such as visual traffic detection and visual landing guidance.”

For example, with traffic avoidance, while a person can immediately identify if there is an airplane in the picture, your typical AI can’t. Well, not without plenty of training. Mora explained that Daedalean uses machine learning technology and vast amounts of samples of right and wrong to teach its system how to accurately identify the “aircraft in the picture.” The process is called “training the model.” When the resulting model is capable of consistently answering the question with accuracy equivalent to a human in the model, it can be implemented as a software application.

What AI isn’t, Mora said, is an intellect making decisions. “It does not possess opinions and does not have any free will. It’s just a piece of software capable of exactly one function for which it was created. For example, analysing some specific flow of data – pixel images from cameras – and then giving the answer it was trained for: There’s an airplane out there.”

PUTTING THE AI IN UAS

As you might guess, while AI’s introduction into piloted aircraft will be more of an “if-AI -complements-the-application” basis, the technology’s range of enhanced situational awareness capabilities will be required for unmanned aerial systems.

“Initial AI routines will generally be focused on increasing automation including elements like 1-Click or otherwise simple/automated launch routines, deployment, mission planning, prepackaged response routines, behind-the-scenes AI enablement of improved data ingestion and analysis, and so on,” explained Don Mathis, CEO at Virginia-based Echelon AI. “AI enables greater situational awareness, and this, in turn, will work with other capabilities like radar to better enable beyondvisual-line-of-sight (BVLOS) operations.”

While the ability of sophisticated drones to operate autonomously BVLOS is the industry’s “Holy Grail,” Mathis says achieving it means solving multiple challenges. “Beyond the regulatory/certification challenges, there is also the form factor and aerodynamic challenges of drones themselves,” he explained. “In the tech world, stakeholders expect exponential growth; in aerospace, we are more limited by physics. So, for example, there are trade-offs in payload capacity.”

“This, in turn, means that processor size (and capability) is relatively limited in a drone if one wants the drone to have sufficient battery power/aloft time and/or have the ability to carry the right kind of payloads (e.g., sensors, or first aid materials) for the mission,” Mathis added. “This means the AI has to be capable of operating on the edge, and this is definitely a more challenging problem to solve than simply writing a generative AI routine in many other contexts.”

AUTONOMY ISN’T AI

Another popular misconception related to autonomous aircraft is that they will all rely on AI. That’s not always the case. In fact, California-based Wisk Aero isn’t currently making AI an element in its family of autonomous passenger eVTOLs, which are currently under development.

“What we are doing is autonomy, and autonomy is not AI,” explained Jon Lovegren, chief of autonomy and airspace integration at Wisk Aero. “There’s a lot of understandable confusion about it. For us, autonomy is about the aircraft’s capability to aviate and navigate safely. We’re doing the same things that a pilot does today, but without having the pilot on board.”

“Autonomy is the capability you can build through rigorous design, redundant architecture, and leveraging existing technologies in aviation (autopilots, TCAS, weather radar, GPS, etc) and taking them to the next level of self-operation,” Lovegren said. “AI is a technology that’s used as a tool to enable certain applications, and we don’t rely on this tool to achieve our level of autonomy. The challenge with AI is proving it will do the right thing every time in a safety-critical scenario – which is required for FAA certification.”

Lovegren said Wisk Aero is “increasing the automation to the point where the operator – who is still part of our system – doesn’t actually have to be in the aircraft. To enable that, we have a much higher level of automation to handle those safety-critical operations. We also have contingencies so if something were to go wrong, the aircraft will be self-monitoring and automatically take the correct actions to ensure passenger safety.”

POTENTIAL GAME-CHANGERS

While the promise of AI’s abilities to lessen pilot workloads is still a work in progress, implementing advanced technologies to help optimise day-to-day flight, ground, manufacturing and maintenance operations seems to be much more readily achievable.

“There are massive economic benefits. We will see highly optimized routes and schedules resulting in greater fuel consumption efficiencies,” explained Yemaya Bordain, Ph.D., Daedalean’s President of the Americas. “Operators will be able to optimize fleet management, making maintenance scheduling more efficient. And many menial operations will be completed automatically, which will reduce workforce costs.”

“Broadly speaking, from an operational and maintenance perspective, it’s going to automate mundane tasks and give us the ability to make sense of enormous data sets and run them through natural language processing to generate logbook entry timelines and work-performed summaries in ways that didn’t exist before,” said Roberto Guerrieri, CEO and co-founder of Arizona-based Bluetail. “For example, things like completing all the paperwork necessary to return an aircraft to service. It’s going to take hundreds of boxes and filing cabinets full of an aircraft’s back-to-birth records and funnel them down to presenting its operational history in a quick and meaningful way.”

“AI could even help combine maintenance schedules and logbooks to quickly identify any discrepancies or what’s coming up on the next task list,” Guerrieri said. “Shifting from manually searching for information to quickly understanding the requirements of the aircraft’s next maintenance phase in a fraction of a second is going to be a real game-changer.”

A MATTER OF TRUST

With all this being said, will AI ever live up to the hype and be welcomed as a valued member of business aviation flight operations?

Experts say we won’t know that for a long time. “We expect to see a slow and careful adoption,” predicted Daedalean CEO Luuk van Dijk, Ph.D. “Pilots may not trust that the new technologies will be safer. They may expect these technologies to add to their informational overload instead of simplifying flying. And, of course, they may fear being replaced by AI.”