Autonomous or Not: Thoughts From Revolution.Aero’s November Town Hall

Last week I had the opportunity to participate in Revolution.Aero’s Town Hall meeting, a monthly event series where hot topics in aviation are discussed by expert participants (and often challenged by brilliant audience members!). This month’s topic, The Autonomous Aviation Revolution, offered opportunities for each expert to share insight into how their companies are tackling autonomous technology, along with their opinions on how it is and should continue to be addressed in the industry.

Early on, the conversation quickly centered around a few key topics: autonomous versus piloted flight, the ongoing prioritization of safety in our industry, and some discussion of future propulsion solutions. While I have many insights of my own on each topic, I want to highlight how some of my industry colleagues are viewing them as well.

Autonomous vs. piloted flight

Despite the title of this month’s town hall, much of the discussion centered around whether or not companies should focus on the creation of fully autonomous technology or if single-piloted operations are a more realistic target. Concerns stemmed around the following:

Timing: Marc Ausman, CEO and Co-Founder of Airflow.aero, noted that the time frame to deploy autonomous technology broadly in urban areas is uncertain, which is why his company has chosen to build a business with a piloted aircraft. Safety: Alternatively, Gary Gysin, President and CEO of Wisk, noted that safety is a primary reason that his company is focused on autonomous flight, citing that over 80% of aviation accidents happen due to human error.

As with many conversations on autonomous versus piloted flight, the question of whether a machine will ever harness the same capabilities as a human pilot was considered. Luuk van Dijk of Daedalean AI believes there is a great deal of misunderstanding and preconceptions about the ability to make machines “smart enough” for autonomy. He mentioned that in avionics, any uncertainty in the environment is currently passed to the pilot. To create an automated system that can deal with real world situations, the data that you train your system on and test must be representative of the randomness you’ll inevitably encounter in-flight. He notes that while machine learning systems are not perfect, they are analyzable, allowing manageable risk if addressed correctly.

I believe that while it’s easy to claim that a system can be safer than a pilot, edge cases where systems are designed on the ground often tend to face challenges. The thought process and the judgement of a pilot is very difficult to capture and model perfectly (as a pilot myself I can say this with confidence), which is why these are the areas that we are concentrated on. Perhaps an ideal solution is the merging of the pilot and the AI system - as we integrate technology into the cockpit, both can learn from one another.

Safety: Regulation and prioritization

A second focus of the panel was safety as it related to regulation and prioritization. Not only do we have to prove to regulators that the new autonomous systems we create are safe, we must also help them understand the support that is required to move them forward. Luuk van Dijk noted that while regulators have been given the task of keeping the public safe, those creating the technologies need to prioritize safety to prevent major repercussions down the line.

At Airbus, our approach is to focus on safety at every turn, which we believe will facilitate autonomous flight down the road. Right now we’re looking at the capabilities and opportunities of single pilot operations; our approach is very methodical and regulatory scoped.

We also have to prove (and often convince) the public that any self or single-piloted systems are safe. Gary spoke to market adoption and how consumers will adapt, mentioning that some will be more trusting to start, while others will test slowly before jumping in. Regardless of a passenger’s risk profile, though, there will always be a subset of the population that has a commute issue to solve for (whether it be getting to work faster or getting home to see their family), and those are the ones eagerly awaiting scalable, autonomous solutions.

While I’m always eager to participate in opportunities that allow me to learn from colleagues, it’s become especially valuable to glean their insights through an innovation lens since starting at Acubed. While we might be taking different paths to get there, we’re all working towards a shared goal - to change aviation for the better.

- Paul Smith