Slashdot Mirror
Can AI-Controlled Cameras Replace An Air Traffic Control Tower? (bloomberg.com)
London's Heathrow airport is testing a surprising new system involving high-definition cameras, monitoring 50,000 arrivals in the month of March. Bloomberg reports:
Views from the cameras will be fed into an artificial intelligence platform from Canada's Searidge Technologies Inc. that will interpret the images and reveal to controllers when a particular aircraft has cleared the runway, allowing them to clear the next flight to come in to land... If successful, the system will initially be deployed when Heathrow's 285-foot control tower is shrouded in cloud -- a situation that currently compels the airport to rely on radar readings to determine the position of jets. That in turn requires a bigger gap between flights, costing the hub nine landings an hour or 20 percent of the usual total...
The same technology could also control the airport's $22 billion third runway due to open for flights by 2025, removing the need to construct a new control tower to oversee the strip north and west of the existing one. The smaller London City airport is removing its tower altogether and deploying a mast with zoom cameras, allowing flights to be managed from the Swanwick control center more than 80 miles away.
The same technology could also control the airport's $22 billion third runway due to open for flights by 2025, removing the need to construct a new control tower to oversee the strip north and west of the existing one. The smaller London City airport is removing its tower altogether and deploying a mast with zoom cameras, allowing flights to be managed from the Swanwick control center more than 80 miles away.
For the day to day and organization of flights I can see AI being able to handle that brilliantly. For more unusual issues I'd imagine controllers will still be needed though; a lot of the issues that are face are oddball ones. A mayday call for instance can involve anything from having an ambulance waiting in the terminal to having a paramilitary response team on the tarmac with fire and EMT's closeby. More mundane but unusual issues such coordinating reports of hazards to other aircraft (think birds, wildlife, debris on the taxi / runways).
You've also got times with radar and other sensor equipment going down at once; loss of radar was a factor in the 2002 Überlingen mid-air collision (though the fault lay in the traffic controllers breaking company policy and managers tolerating it).
That said, if a computer is able to handle even 20% to 30% of the workload, and can integrate well into the controllers workflow, then it's worth looking at. This leaves more mental focus for controllers to deal with situations that the computer isn't good at.
To tell when a plane has left the runway ?
If the purpose of the tower is to actually see the plane, you can't just put a camera at the end of the runway looking down it and route it to a display ?
AI and computers/robots are being used to replace higher income jobs to increase profits for companies.
When some idiot programmer develops AI applications to develop AI applications, the AI programmers will be on the scrap heap.
The cut even cost for robots replacing telephone sanitisers means their jobs are safe.
Go well
Self flying / taxxing aircraft aren't that difficult a problem and they would interact well with automated air traffic control. As long as there are humans though, I think fully automated traffic control would be difficult. Instructions are still given by voice and they can be misunderstood and / or garbled etc.
Automation is also very poor at dealing with unusual situations - say a major earthquake at SFO that might have damaged runways, or a piece of debris falling off of a plane and creating a hazard, or a confused inexperienced (or sometimes experienced) pilot blundering into the wrong airspace. I think some humans will be needed to deal with these sorts of situations.
Almost sounds to me like you're responding an entirely different question, since the original was about a tower.
Ezekiel 23:20
Answer: Yes.
If you still have humans in the loop, decisions will be delayed when seconds count, and the biggest mistakes will happen when the controllers are exhausted and confused.
The worst aircraft accident in history was the 1977 Tenerife airport disaster in the Canary Islands. 583 people died in a cascade of errors, miscommunication, and poor judgement calls, as exhausted controllers dealt with heavy fog, delayed flights, and frustrated pilots.
One pilot thought he had clearance to takeoff on a runway that another plane was given clearance to cross. Automated cameras would have detected this problem, and could have automatically ordered the planes to stop. This wouldn't even need AI. Just a dumb algorithm would have saved hundreds of lives.
Can AI-Controlled Cameras Replace An Air Traffic Control Tower?
Absolutely. Sure flying might become Russian roulette but hey, you did replace the ATCs didn't you? ;)
Anons need not reply. Questions end with a question mark.
No chance in hell Heathrow will open a third runway in 2025. It is unlikely they will even have a ground-breaking by then.
Not only cameras, but monitoring the distance between planes, when they are in position to take off, the engine spool - up and launch, and rate of climb until the plane is out of the area. Likewise in reverse for landings - watch landing speeds, position, etc. Much like self driving cars that need human monitoring, automation at least at this point, can't account for unseen variables and therefor, a human must be monitoring at all times. both flight controller and pilot. It would be possible for AI to make sure the flights are properly spaced and staying within parameters. Yes this would be very valuable for inclement weather takeoffs and landings. Imagine an AI being able to check tarmac conditions to account for water or ice, which affects landing distance, and being able to let the tower and pilot know so they can compensate. I see it as a safety measure more than as traffic management matter
Republican leadership = Idiocracy
If successful, the system will initially be deployed when Heathrow's 285-foot control tower is shrouded in cloud
How can camera work around clouds?
Remote ATC has been in operation for at least 50 years. This is just a natural extension of existing operations. For example: the Edmonton FIR (maybe world's largest) uses remote VHF transceivers to provide direct control of Northern locations. The controller is in Edmonton but talks directly to traffic in the Yukon, NW, and Nunavet Territories.
High definition cameras may provide better operational coverage than a local tower. There is a real opportunity to integrate infrared sensing and ground radar systems to provide useful information even in bad weather. Control of ground operations in a busy airport can be a nightmare with unexpected and asynchronous events causing all sorts of problems. Replacing the glass window and binoculars with screens may reduce environmental distractions and lower controller anxiety. Good information and no blind spots goes a long way to improving safety.
Once the tower controller's view is changed from a window to a screen, it really doesn't matter where the controller sits as long as the communication link is reliable and has low latency. As always, better networks create opportunities for new services.
Remote presence tech is well established. This is no surprise.
I did not get that they are replacing controllers. They are providing additional tools and remote sensing. The system does not provide automated clearances to pilots as far as I understand it.
"Views from the cameras will be fed into an artificial intelligence platform from Canada’s Searidge Technologies Inc. that will interpret the images and reveal to controllers when a particular aircraft has cleared the runway, allowing them to clear the next flight to come in to land."
'nuff said.
Youre right.
The answer is no, and the full article is bullshit.
Anyone who knows a thing on ATC knows cameras are not involved and are not a good idea to locate planes.
Furthermore, current A.I. has absolutely no accountability and traceability on the decisions and guesses it makes, making it totally unsuitable for safety critical applications.
aaaaaaa
"One pilot thought he had clearance to takeoff on a runway that another plane was given clearance to cross. Automated cameras would have detected this problem"
On heavy fog? Doubtful.
As you say, main problem was miscommunication and the after-fact was an update on protocol communications (more strict vocabulary an syntax, mainly).
The problem here is not that technology can't help -it certainly can, but that it's not going to be implemented to increase security but to lower costs: you see, cameras will be deployed so they can save money on another control tower and, you can bet, on tower controllers since now the same people will take responsibility of more runaways. Nice for security, yeah.
"I did not get that they are replacing controllers"
Well, I do so: "The smaller London City airport is removing its tower altogether and deploying a mast with zoom cameras, allowing flights to be managed from the Swanwick control center more than 80 miles away."
They surely will move each and every controller to Swanwick, yeah, sure. And for new deployments, they'll hire just as many controllers as they would do if they were deploying new control towers along new strips, yeah, sure.
Technically possible but ...
People will always want a person/organization to be held accountable when things go wrong.
Any person who will be held accountable will want to ensure some control over that for which they are accountable.
So there will always be a human in the loop as long as people expect there to be a human to blame when things go wrong.
"Consensus" in science is _always_ a political construct.
So some consolidation may occur, what of it. The workload is what it is whether at a local tower location or at a remote ACC.
The point is that this is not AI software replacing humans. Good grief, get a grip.
Furthermore, current A.I. has absolutely no accountability and traceability on the decisions and guesses it makes
A curious claim. Does your definition of "current AI" exclude the automatic reasoning systems with tracing that we've had for decades by now?
Ezekiel 23:20