Originally Posted by
JerryG
Watch this space, mark my words, and then argue with me in five years from now! (More like two years actually)
The idea of a decentralised self-organising ATC is an interesting one. But it is not new – I remember some clever people talking about this over twenty years ago. It kind of went hand-in-hand with the concept of direct or free routing which was also a bit of a Holy Grail back then. The problems are formidable and not necessarily solved by waving the AI wand.
Of course, it is possible in the extreme case where all of the actors (aircraft) are known to one another and are all in communication with each other even before any separation action is required. There will be an optimum routing solution and AI may well be the way to solve it. However, ATC has a nasty habit of introducing variables at frequent and inconvenient times and any “solved in advance” solution based on more than a very small number of actors will go wrong and have to be re-solved dynamically.
For something like drones which can turn on a dime or even stop completely then a dynamic solution may indeed be possible. For large aircraft with high speed, constrained acceleration, deceleration and limited turn rate the problems become much more difficult. Indeed, it is almost impossible to prove that, in a complex environment with high traffic density, an aircraft cannot end up driving into a safe-separation “cul-de-sac” requiring TCAS or some human-intervention (if the human is even still in the loop!) to save the day.
It may be solvable but, except maybe in some very niche scenarios, not in five years.