Air traffic control is an obvious important part of the aviation industry. They control the some 45,000 flights in out of the U.S. each day [UCMerced]. The job of an air traffic controller is known to be super intense, and has been said to be one of the most stressful jobs in the world. With rapid advancements in Artificial Intelligence and the increasing focus of mental health, does AI actually stand a chance at assisting, or even replacing, air traffic controllers?
There are 2 components of AI as it pertains to air traffic control—thought, and speech. The speech portion could easily replace humans as society today consists of automated voices that are spoken in an AI format currently. The “thought” portion is where the question lies. Human air traffic controllers rely on training and experience to make quick decisions that keep planes safe. The critical thinking and procedural decisions that an air traffic controller makes allows so many planes to be in the air at the same time, yet separated, with speeds coordinated where needed.
According to EuroControl, AI can revolutionize ATC by augmenting human capabilities and automating routine tasks. They also state that AI can forecast traffic patterns, weather conditions and potential conflicts, and enable controllers to make more informed decisions. There’s still the human element though. Generally, humans still remain stronger at exercising judgement, making quicker decisions, and are able to handle unforeseen situations.
AI is currently being tested and trialled through SESAR research. The AI ATC is being used for taxiway inspections and runway monitoring. AI tasks in the inspection discipline include bird hazards, presence of drones, autonomous vehicle monitoring, human intrusion, and more. AI can also learn and detect hot spots, and alert human controllers on the need for additional support.
One scenario, hypothesized by Amy Pritchett on The Conversation, indicates a strong need for humans and the reality that AI cannot fully perform the duties of a human air traffic controller. In the presented scenario, a pilot cannot confirm, based on cockpit indicators, if their landing gear is up or down. The next step would potentially be to perform a “low-pass” over the runway. Only a human in the ATC tower would be able to look at the belly of the plane to see if the landing gear is extended or retracted. If it is still retracted, it would require a human to recognize the issue, and take the necessary precautions to alert ground emergency vehicles to be on standby when the plane lands with a gear-up landing.
Dr. Pandey’s of UC Merced performed an Air Traffic Control study utilizing AI. In summary, he concluded that AI certainly can assist and advance the industry. However, he also recognizes the risks that simply cannot be mitigated without the human element involved. There are too many planes, going too many places, with too many human lives at stake, to risk it all on a “computer” performing all of the work.