Artificial Intelligence startups now focusing on new forms of problem solving may be coming to the aviation world far sooner than anyone expected. For years the public has explored the implications of artificial intelligence on aviation and aerospace when it comes to how aircraft fly, how we monitor those flights, and assessing who is doing it with significant success.. What hasn’t been discussed quite as much are the ways new tools – driven by the maturation of multiple technologies – are impacting the use, implementation, or connectivity of the aviation industry. New problem-solving techniques that address not the flight of, but the data captured by autonomous vehicles in the sky, on the ground, or under the water is immense. As most in the UAS industry will tell you, the drone itself doesn’t matter – it is what the drone is doing that matters; and if new technologies make a drone more effective you can be certain that it will matter as much as new forms of control or communication. The same goes for any technology part of the data capture and delivery chain – satellite, drone, train, boat, car, scooter, camera, secret agent, etc. Increasing the value of the data captured will have a direct effect of making the industry more successful; without that value the growth will slow.

Take for instance drone inspections, for a decade or more the power generation, distribution, and transmission companies around the world have tried, and in great success, to implement drone technologies into their systems. Industry leaders and early adopters like Puget Sound Energy, San Diego Gas and Electric, and Dominion Energy have been able to integrate drone data capture into their operations in a way that saves lives, reduces maintenance down-time, and in-turn saves costs that may otherwise have been passed on to customers. Drones, often DJI or Skydio systems, capture immense amounts of data that is then passed on to offices within the organization (or sometimes contracted out) to analyze, categorize, search, and save (for different periods of time depending on privacy needs and usefulness of the data). However, even from its beginnings the amounts of data being generated were enormous and often too large to fully take advantage of. Some programs were even scrapped or downsized simply because the systems in place to absorb and analyze data could not handle what was being delivered. Ultimately, it was not safety of the system, passion of the people, or rigor of the regulations that hindered adoption but the inability to integrate the data in a way that was useful.

Today, we’ve seen an up-tick in drone data collection organizations implementing the technology likewise grow, and maturity in data collection and data hygiene enable significant cross-systems integration. The systems themselves are now evolving to meet the needs of any size organization and are now able to connect across technologies in a collection agnostic way. What would have been only a dream for data scientists looking at terrabytes of data and being asked for specific pole numbers in maintenance images, will soon be able to leverage the forms of artificial intelligence being explored via ChatGPT or Dall-E. One example of this is the application of generative AI to data mining, sorting, and analysis but it really serves just as an example for the tremendous impacts that these “non-flight” related evolutions are having on advanced aviation.

So, I’ve heard the buzzwords but what is AI?

First, I don’t claim to be an expert in all things AI, but let’s give it a shot to explain the basics of this new form of problem solving. In its simplest form, Generative AI is a set of algorithms designed to “learn from training data that includes examples of its desired output.” To create an output the algorithm will take any guidance parameters you’ve given it, search for patterns and structures within the data it has been trained on and generate an output. The more “good data examples” you have to train the algorithm, the better the returned output will be in answering whatever prompt you may have.

So how does this relate to aerospace? Well, we’re now seeing the first real implementations of this new type of technology with the recent emergence of companies like Danti, designated as “the market’s first Earth data search engine that enables expert and non-expert users alike to pose simple questions about physical places on our planet and instantly gain access to the breadth of information generated daily by satellites, drones, analytics firms, social media and the like.” Simply put, an average user who needs to answer a complex question from a geospatial perspective – a question that might otherwise have taken weeks to accurately solve given status quo data sifting and analysis techniques across platforms – can find the results they need in seconds. While not an aviation tool per say, it makes the value of drone flights significantly greater. Danti will be applying machine learning and natural language processing techniques to enable geo-spatial data from a variety of resources easy-to-access and easy to gain insights from across the entirety of connected data.

By making data useable for the average person, or the average non-expert analyst, we can all make better decisions and be informed faster, cheaper, and more reliably. While it may not be as flashy as self-flying drones or advanced air mobility passenger aircraft flying above the Paris Olympics, it is often technologies like these that ensure that drones, aircraft, and satellites work together to be useful that drive industry success and adoption.

A similar evolution took place just about 10 years ago that drove the rapid evolution and adoption of a new technology in aviation. Just a decade ago, we saw the integration of flight controllers, GPS, and cameras to create the first consumer focused drones. If you ever flew the old-school drones prior to the advent of flight controllers that would allow for stationary positioning and self-balancing you’d know that it took immense skill, experience, and lots of technical knowledge (for rebuilding after crashes). With the implementation of advanced flight controllers and intuitive AI, any person could quickly pick-up a quadcopter and fly or even start taking professional grade videos. This evolution democratized in many ways the low-altitude skies and in the same way, the implementation of Generative AI may reduce access cost for meaningful outcomes.

So, when we ask, “Will new forms of AI begin to drive value in the drone industry?” The only answer can be yes, and in ways we may not have thought about.