Will ATC ever be automated?

[Quickstatus]

ATC is likely to be the profession least likely to be automated.
Comments?

 

[Hvr]

Not successfully or safely.

 

[moa999]

Wouldn't be the bottom of my list.

Highly manual things like construction trades, animal butchery etc would be well ahead.

Indeed I suspect an ATC controller from 50yrs ago would struggle to walk into the job today given the amount of technology used, compare that to a butcher or carpenter.

 

[SydneySwan]

This is a question for DT. He is an expert in this area.

 

[Forg]

I reckon yes, but it will be a VERY long time … technology seems to be very conservative in the aircraft/flight industry, and I kinda think I want it that way!

 

[harvyk]

Parts yes, but never fully (at least not anywhere in the near future).

AI is both very smart and very dumb at the same time, and in the ATC world it's not just a case of mis-counting the number of r's in strawberry, or adding 8 fingers to a hand, it is lives on the line.

Computers are really good at pattern matching even across very large amounts of data (that's effectively all AI is), but it's not so good when it comes to thinking when an unusual situation occurs.

The most likely scenario I see is one that computers / AI allows human controllers to control much larger areas, thus reducing the required headcount. Automating some of the routine tasks.

So yes, I do see a future where AI augments human controllers to make their jobs easier, but I see humans in the mix for a long time yet.

 

[jb747]

It will happen, in the same time frame that all the passengers are also replaced by AI.

 

[elanshin]

ATC is likely to be the profession least likely to be automated.
Comments?

I believe the technology to allow for this will happen. But the human aspect of it will not.

Basically, having computers be able to map in real time routes, plane positions and instructions should be technologically possible at some point. But the psychological human aspects (like self driving cars for example) and trusting it will take a long time to develop.

What is more likely to happen is augmentation. Think of it like autopilot where if everything is cruisy and nothing strange is happening, there's some levels of automation to help ATC process the information and even instructions - especially if we're having ATC staff crunch.

Although never say never. Automatic lifts were considered an impossibility at some stage of time, but no-one would even consider it now getting into a lift.

 

[SCM]

I think automating ATC would be very easy, but the problem is what happens if there is any failure at all, how do you provide a safe backup? So we're back to a fully staffed human ATC, thus there's not really any point in automating it.

What may happen is that an AI will look over all communications and intervene to stop and problems before they become catastrophic.

 

[elanshin]

I think automating ATC would be very easy, but the problem is what happens if there is any failure at all, how do you provide a safe backup? So we're back to a fully staffed human ATC, thus there's not really any point in automating it.

What may happen is that an AI will look over all communications and intervene to stop and problems before they become catastrophic.

If this was properly invented and a failure occurs, there would generally be enough people still supervising and managing (also for emergencies) to stop all traffic taking off and directing the existing traffic. You mught have a few planes put into holding patterns and the staff working their cough off but it wouldn't necessarily be unmitigated disasters.

But if we take yesterday as an example, one sick ATC supposedly caused the fiasco yesterday at SYD, so it's not like human systems are without fail points either.

 

[Doctore1003]

it's not like human systems are without fail points either.

Yes, people often overlook this point. Once we have an AI system that performs better than humans, wouldn’t it be negligent to use humans?

The more likely roadblock to AI adoption may be political pressure from irrational voters.

 

[elanshin]

Yes, people often overlook this point. Once we have an AI system that performs better than humans, wouldn’t it be negligent to use humans?

The more likely roadblock to AI adoption may be political pressure from irrational voters.

The problem is acceptance which is the same problem as self driving cars. Statistically i believe they have already reached a point where they are better than an average human driver. But we as humans can't accept 70-80-90% success rate even if our own is lower.

The barrier to acceptance is likely well in the 99% rate which no human will ever achieve. But that is just human psychology.

Overall, ATC automation will happen, but likely more like autopilot systems where there are backups and human hybrid input. I do not believe there can be any doubt that a computer can track and maintain in 3d space multiple airborne vehicles with loads of different parameters better than us humans ever could. We as humans just aren't built for parallel processing - we can train for it but ultimately will never be better than a computer at it.

 

[Quickstatus]

Has anything in ATC been automated/computerised

ATC can be local at the airfield , or FAB , continent

 

[elanshin]

What aspects can be automated?

Instructions for one in general when theres nothing crazy going on. Computer can easily track everything in 3D space and give instructions based on pre approved flight plans for vectoring in busy "controlled" airspace with people monitoring the situtation.

for example.
Controller says something like: "QF1 Sydney Departure, Radar contact. Initiating automated controls. Good day." Controller presses a button and it's now in "automated" mode and a computer sends instructions periodically as needed.

Let's be honest half of the work there is very routine and can be automated. Controller can monitor the situation and more airlines. Any major deviations gets tagged for the controller to follow up on. In theory this could all be fully automated but that has its own risks with human backups not being as alert so there should at least be some level of human contact.

 

[markis10]

Interesting question with no easy answers.
How many have expressed frustration with voice recognition phone systems, now think of them in an ATC context!
Can AI detect issues with a read back that hint the message has not go through?
What resilience does a system have when computers fail (crowdstrike anyone), how easily can a back up non automated system kick in, and how secure are the systems in the first place?

Given self driving cars are yet to be commonplace yet run in a 2D world, 3D ATC automation seems sometime away.