Disclaimer: The views on this page represent my thoughts not the facts I know! For examples of how little I know please go to the questions page."
Why are people trying to do a computer’s job and control the flight paths of planes in our skies?
This is one of the questions that I have always wondered when I watch films and documentaries about air-traffic control. Is it that people are just significantly better at spatial awareness? Are there too many different combinations of flight paths that mean a computer cannot work out a solution fast/optimal enough?
To these questions (and the many others that I can think of) I see the air-traffic controller’s job under serious threat of extinction. The skies are filling up with planes and I do not see how they can keep up. Yes, people are good at spatial awareness, but up to what limit? Twenty planes? Fifty planes? With every extra plane comes a greater chance that something will be missed. A person can only check so many flight paths against the rest and when that limit is reached, errors happen. This is where a computer is in its prime, checking and re-checking the flight paths.
Now for the computational explosion. The problem is to find optimal flight paths for all the different planes, when there are an infinite number of different possible flight paths each plane can take. This makes for a very difficult problem.
Can a computer really not solve this problem? The real question is, can a computer find the same level of optimal solutions that people can work out? The first thing to realise is that people cannot possibly think of all the different solutions, we have intuition, common sense and strategies to minimise spatial overload. I think solutions people come up with will be good, but not optimally brilliant. I am sure that air traffic controllers do think, “Oh, if I’d have done that differently it would have been better”. So, the computer hasn’t got to get the optimal solution, just one that competes with the human one.
Is the computational explosion really so big? For example when an air traffic controller tells a plane to turn to a heading they don’t say “turn to 272.3125” i.e. there are no fractions involved just whole numbers, this alone should dramatically cut down the number of options a flight path has! Also, the time at which the command is given and the reaction time of the pilot will all mean an inconsistency between the computer solution and the actual events. Whilst searching for the shortest flight paths, a sub-criteria could be to search for the maximum distance between the planes at all times. This is where a computer could be better because you can have multiple criteria, rather than just searching for the shortest flight path.
The solution I see for the future is where the computer makes the decisions (maybe overseen by people) and people are employed to communicate with the pilots. You could have 4 or 5 people with a screen and headset and the computer displaying a flight path for a plane, what the next turn will be and a countdown to when the person should contact the pilot to get him to turn to the next direction, adjust his speed and height. Once the “controller” has seen that the plane has turned, they click the confirm button and a new plane’s flight path is displayed for turning and the countdown begins again.