Right now, air traffic controllers are asking themselves the same question as everyone else: will my job ultimately be replaced by AI? As with other forward-thinking sectors, the ideal approach for everyone will be to combine the strength of both machines and human operators.
But how will AI come to shape the air traffic control industry? How can AI augment the capabilities of experienced ATC operators, leading to the safer, more efficient management of airportsâ skies and runways?
Key areas of AI/human collaboration in ATC operations
Collision avoidance: The top priority of any ATC operator is to avoid collisions between planes in the air or on the ground. However, the process of assessing conditions and delivering instructions necessary to prevent such a disaster is entirely human-controlled, hence it is slow. Once they have learned sufficiently from historical and operational data, AI platforms can speed this up by predicting and determining safe routes and planned instructions faster than even the most accomplished ATC operator could manage. Human oversight will remain critical, however, as both final decisionmaker and failsafe mechanism.
Conditions updates: Even in routine, non-emergency operational circumstances, AI analysis will become increasingly useful. It can keep ATC human staff appraised of external factors, such as changing weather conditions and airspace interference from outside organisations, in almost real time. Giving human operators more operational visibility and the tools to enact decisions more quickly will be game-changing upgrades to their overall effectiveness.
Flight route planning: Alongside the quick reactions and decisions, the larger strategic picture of organising hundreds if not thousands of daily flight movements cannot be ignored. Human ATC operators are still responsible for the plotting and ordering of these routes â a task that is extremely time-consuming and often error-prone. AI will take the sting out of the complex strategic airspace planning piece, driving efficiency and reducing the instance rate of costly or even dangerous mistakes.
Exemplar project: Project Bluebird
Project Bluebird is a NATS/Alan Turing Institute collaboration effort with the ambitious aim of delivering the worldâs first AI system capable of controlling a section of airspace in live trials. Set in the UK (which has some of the worldâs busiest airspace), this will be an industry-changing project with ramifications for the entire ATC sector.
Once fully developed, Project Bluebird should be able to generate a probabilistic digital twin of UK airspace. A real-time, physics-based computer model will predict future flight trajectories, taking into account all relevant conditions changes (weather, etc) so that it can inform operators exactly when and where it expects every tracked flight to be at any given time.
Using this digital twinning approach, combined with millions of historical flight datasets, the AI will be able to collaborate with human operators to pre-emptively detect and avoid high-risk aircraft conflicts. At the same time, it will keep updating the strategic planning of all flight routes âon the flyâ. Unlike traditional human-centric decision-making, this AI-driven approach can stay ahead of changing conditions and optimise flight paths before delays and conflicts have a chance to develop.
Equally importantly, the AI will learn from its human partners. Project Bluebird envisages the use of human ATC operators as instructors to teach the AI about ethical considerations and more instinctive decision-making processes. The ultimate aim is to enable the AI to balance safety and efficiency, and think like a human by going beyond the immediate data whenever it becomes necessary.
AI is coming to ATC, but weâll have to be patient
The scale of resurgent airport operations in a Covid-recovering aviation industry means that flight movements are going to reach levels unmanageable by human operators if they remain unaided. The need for augmented ATC operations is clear and the risks are increasing with every added flight making the skies more crowded. Already, AI-based solutions from trusted providers like SAE exhibitor Air Traffic Solutions are impacting the market and driving safer and more sustainable airport operations.
However, the testing of AI cannot and should not be rushed. Comparisons can be drawn to the automotive industry, where self-driving cars and networked roads are still decades away from being commonplace due to the need for exhaustive testing. Leaving the fate of multiple flights to the decision-making capabilities of an AI platform â even with human oversight â will mean even more stringent testing and failsafe arrangements than anything travelling by road.
