I'm all for digital air towers and improvements, but decentralizing air traffic control into one location instead of local air traffic controls at each airport seems like a great way for one internet connection to ruin 20 airports
London City has a triple connection. 2 independent optical links that run separate routes and if both fail, a wireless link. And you don't need many controllers for airports that only see a few aircraft a day.
@@LeeeroyJenkinsit’s not really replacing people. You still need people in the inner workings and stand by. People are still there doing the job so to speak. It’s just that the use of AI streamlines and automates the process to make it more efficient and possibly safer.
As an airline pilot I can say that for sure at least in the US, the ATC system is under extreme pressure. In 20+ years of flying I've never heard the frequencies be so congested and the controllers sounding so winded at peak hours, and they often have to resort to many delay vectors even on perfectly clear days. It's not their fault, just the nature of the beast. The FAA innovates at the speed of molasses due to all the regulatory red tape involved which is great for safety but terrible for efficiency. In the early 2000s they came out with the new "nextgen airspace" moniker and promised very lofty things to be in place by 2020. It's now 4 years past that and we've hardly seen any of that come to fruition, aside from ADS-B which while great has also has placed quite the burden on many other areas like GA pilots, drones and RC hobbyists, all of which I'm also involved in. Some smaller, currently untowered airports would certainly benefit from this technology, but it's still far far away from being viable for any of the top 100 busiest airports in the world. I've dabbled plenty with chatgpt and Gemini and as impressive as they can be they're extremely stubborn and rigid at times, and seem completely unable to improvise in some complex situations. This could be a life and death thing in an emergency when airborne.
As a AI scientist, I agree with you. A lot of the machine learning technology are not mature enough. The way they are trained makes it impossible to act on their own in new situations that have been not been encountered by these models. I wouldn't feel safe if they deployed these models to safety critical applications like ATC. It's way too early.
No existing ATC has anything to worry about during their careers. But I remember having a conversation with a tower controller once (I’m a center controller) and somehow we ended up talking about why I wouldn’t recommend anyone new to go to the tower route. Among many reasons, I also said how the towers are going to be the first to go. It will start with the small towers first until only the Core 30 towers are left.
an air traffic control tower is often the icon of an airport anyway, i think for major airports the digital tower would probably be a supplement to real towers i only see purely digital towers being in use in remote airports, where they get maybe half a dozen flights a day, because the number of flights are so low, a tower out of order wont be detrimental to safety
Not being in the tower and able to look with your own eyes seems like asking for a digital disaster to happen with no immediate work-around. Cyber attack, power-outage and the backup fails at some point… someone’s gonna be sent running up that decommissioned tower and they’re not going to be in shape to do it.
At the end of the video, they showed an ATC moving a window of one portion of the airport to another section of the screen, essentially moving a taxiway virtually. I think this type of thing can only lead to confusion and deadly accidents as the ATC lose track of the relative relationship between aircraft and airport locations. Seems like the whole point of those giant screens is to display the entire airfield at once and maintain situational awareness. The option to transpose taxiways is dangerous.
4:53 I'm a pilot in Colorado and flying into Northern Colorado (FNL) was always interesting. It's treated like a Class D airport while not technically towered. ATC is on-site in trailers with a camera pole and they'll request you report X miles out since they don't have radar.
Im sure AI assisted with human controllers would be okay. Like an extra set of digital eyes cant hurt. But still need humans as backups. Tech is prone to mistakes, crashes, errors, that sometimes arent fixed immediately like an airport would need. Id say its a fine line between mixing the two
@@electrology Accountable is an adversarial word. We learned long ago that flight incidents are caused by a chain of events. The entire chain must be discovered and analyzed. Not to assign accountability, but to take measures to prevent the chain of events from happening again.
I can see virtual towers on site where a normal tower sits, but it is madness to have virtual towers miles away... network connectivity is far from a sure thing, even when multiple redundancies are in place.
You might could read up on how NORAD communicates with America's armed forces clear across the world. I don't think America will introduce this concept unless it's better in many many ways.
@@DinoAlberiniIn the IT and telecom world, "guaranteed level of service" means that if they don't meet that level of service, they discount the fee they're charging you. It doesn't mean it's impossible to have an outage.
@@DinoAlberini Redundancy quite literally means relying on multiple systems to reduce the impact of failure. If a centralised ATC service fails, the entire country loses the capability to direct traffic into and out of airports. If you have a team at each airport, failure means just one airport goes offline. And nothing is quite as robust as being able to look out the window.
Makes sense. I know there's a lot of dependence upon electronics for aviation. But even still, what happens when the power is cutoff? Being able to see it yourself seems like an excellent backup plan, and would require someone be there to see what's going on.
As a 16 year veteran controller in the USA I can admit our system is saturated and frequencies are packed in the busier air spaces. And I hate to think that I can not be replaced but at the end of the day, the pavement at the airports can only handle so many planes within a limited time. We get so efficient and introduce more aircraft and sort of get back to the original delay. I’m curious to see where this goes! AI is rapidly improving.
I think AI is going to rock your world the way computers improved productivity when they were introduced. Safety first. The railroads use to mandate a caboose with a live operator in them in early trains. Then they went without operator because of technology then they did away with caboose. LOL
THE SAME REASON THIS IS A BAD IDEA is the same reason it is a BAD IDEA to paint all your car windows with black paint, attach cameras all over the car, then installing monitors in the car, and then taking the car out for a drive near a school zone when kids are getting out of school. If you find that scary, relate that same scenario to airplanes.
I dont think air tower control will disappear. You always need someway to do things manually. If those system are in maintenance or fail. You need something to do by the eye.
What happens when a bird lands near the tower, or bad weather blocks the camera? What about when the power goes out? I can't be the only one thinking about this.
If bad weather blocks the cameras, it would also block the tower from seeing too and if the power goes out at the airport then so does the towers power at the airport But if you have multiple Digital control centres and one goes down, the others can take over their airport operations
You're not. This is tech companies trying to sell technology that doesn't really work to unsuspecting officials. I've seen this countless times in manufacturing. Automatically guided carts that 1) couldn't avoid people 2) couldn't emergency stop 3) worked way below the projected line speed.
"What happens when a bird lands near the tower" I'd assume multiple cameras in different position with overlapping fields of view would solve that. "bad weather blocks the camera?" Bad weather (heavy fog, rain or snow) will also blind the controllers looking from the ATC tower. I work at an airport and there's a reason why LVP (Low Visibility Protocol) is a thing and every emplyee knows what that means: different rules for going about your work. "What about when the power goes out?" Airports all ready have their own backup power generators for just that eventuality. And if proper safety and control systems can't be brought back up quickly enough the planes will divert to a backup airport. "I can't be the only one thinking about this." No you are not and before anythng like this is actually adobted it will go trough extrimely rutheless testing. There are international and national authorities that will need to accept this kind of tech before it'll be accepted.
Low visibility already causes accidents on airports even now. Non-visual sensors can ease this issue by providing alternative means of information. This is a plus. And if power outage means no screen is working, it also means no radio is working. So no one would be dare to install any vital electronics without an alternative power source. These are not problems unique to smart technology.
What happens when there is an outage? Or perhaps a cyber security breach? It would be much safer to use line of sight from a tower and standard radio communications to control the traffic. This seems ripe for failure. Sometimes things should stay the same. Not to say we can’t empower controllers to do an amazing job by empowering them with technology but removing them from the airport and staging 70 miles away is just asking for problems.
What happens when it’s foggy? Or if someone comes to work drunk? Humans are just as fallible than computer systems. It’s about costs and capabilities and this video makes it clear there’s humans manning ATC functionality, they just aren’t using binoculars, squinting, and coms to figure out what’s going on…
What they have done in UK and elsewhere is designed a system where controllers can be remote workers. Can the controller be given computers and displays so they can work from home in their pajamas?
It’ll be interesting to see how quantum radar (which is 65,000 more sensitive than regular radar) will be integrated into Air Traffic Control (ATC) systems …
i think that this is great but... if any power outage happens you can't see the planes... so, this should be an aid NOT a replacement. Eventually the tower will be a watch area but the main comms can happen in the digital room
Yeah… a remote station works great until some construction worker accidentally smucks a fiber line when doing some ditch maintenance. Hope they have numerous connections ready to go for backups.
Especially if they can get multi-spectral cameras (infrared, etc.) in on the act, they'll have better visibility, WRT what is where, during adverse weather conditions. The would enhance safety, specially in nighttime, fog, etc. The next thing they need to add, and the NextGen system in the USA is supposed to provide, is digital text information between ATC computers and pilots. A pilot could notify the ATC system that, for various reasons, they can't make their takeoff slot. That open slot would be notified to all other aircraft, such that another aircraft could take that slot (such as the next in the queue, assuming they'll be ready in time) and things can move along more smoothly. Meanwhile, as various aircraft move up, a slot would open later in the queue, such that the earlier aircraft can still get one in a timely fashion. This could all be handled by computers, without needing human ATC operators getting involved. Scheduling the various flight out, queuing planes onto the taxiways, etc. could all be handled to greater degree by computers and the pilots, reducing the ATC operators' workloads. So long as digital networking tech is performant (enough bandwidth and sufficiently-low latency) and reliable, it should be possible to put the actual ATC operators just about anywhere. It might become possible to have a handful of ATC centers in an entire country (as demo'ed in Norway) coordinating all the major airports. But there would need to be redundant networking; working for internet service providers over the years, I've been amazed how many times a "cable-seeking backhoe" managed to find the actual cable providing the company's internet backbone and take them completely off the net. Having a pair of different backbone providers, through different cables, coming in through different physical routes, was the only way to avoid that.
You’re describing the ATFM, air traffic flow management. As far as infrared, that’s ubercool and I’m totally for it but radar and ADS-B are all they’ve given us.
When the NATs system works it’s all well and good, but when it fails it’s a right mess, as it did on August 28th 2023. The incident was triggered by the inability of the Nats computer system to process unusual, but correct, data in a flight plan submitted for a plane crossing UK airspace to Paris from Los Angeles. Two waypoints with an identical code triggered a “critical exception error” that caused the primary and secondary systems to disconnect themselves and stop working. On-site technicians couldn’t resolve the issue, and the specialist engineers were working from home. It took 90 mins for an engineer to arrive on-site, and it took a further four and a half hours till the fault was resolved and UK airspace could operate at normal capacity. But the damage was done, due to safety regulations restricting staff flying hours, planes being in the wrong place, and other subsequent logistical issues, over 3000 flights in/out of the country were cancelled, 700 000 passenger’s stranded and disruption for a 3 days. A real utter calamity, and that’s the risk you run with centralised operations, yes they are more efficient, cheaper to run, but when things go South, the negative effects are more far reaching and impactful.
There is so much the FAA could do Let’s start with PDC, D- ATIS how hard can that be I can see it at smaller GA airports with high traffic But how do you keep cameras operational unless you have lots of redundancy
It’s a great idea as long as everything works perfectly. If you have say a contractor that wipes out the fiber optic cable to the master control facility then there might be some problems. If there are problems with the radar or ADSB transponder then there might be problems. If the local radar or ground surveillance radar fail there might be problems.
For better visibility, identification, and logistics - great idea! But I'd never fly again if actual air or traffic ground coordination was controlled by AI or any computer. A computer can identify hazards and alert the ATC, but shouldn't make those decisions on its own. Also: want to cause some trouble? Hack one of those remote tower camera feeds and watch the economy crumble
How cool and nice this looks all the AI and digital distance control. this AI/Digital tower operation is a very big risk and target too for hackers and digital warefare, they better invest good in digital security on all layers of use and training for people how to work the system safe and secure !
some of the best engineers in the world are working on this, don't you think they'd think of that and have redundancies? your life for the past 20 years has been run via servers, how often do you see them go down?
So what happens during a power outage? internet outage? or damaged camera/ computer? how will airplanes get to the ground if the only have 15 mins and all of those happened all at once?
I think this move needs some strong pros and cons analysis. Yes shifting to full-automation will make analysis and planning more accurate and quick, but we'd be over relying on 2 things: electricity and IT. Once either of the two has a problem, there's almost nothing anyone can do in case of an emergency..?
Agreed, IT issues have grounded flights in the past. We should always have backup systems for IT failures. Computer vision isn't 100% reliable either. I'm guessing this "digital tower" thing is a push to cut costs on tower construction and to outsource air traffic control jobs.
2:38 Hearing the CSO of the system refer to the AI as “monitoring a wide range of things simultaneously, whereas a human has to focus on particular areas” and “omnipresent” is extremely worrying! It shows both a near-total lack of understanding of how it works and a massive amount of complacency (ironically, a lack of focus on the danger). Those are two of the main ingredients for air industry disasters. The entire basis for the tech they call “AI” in these systems is that it focuses similar to a human, and lacks all forms of short and long-term memory (the “keep in mind” context of a situation). For example, an AI model will “happily” place a callsign tag on one plane crossing a runway and another tag on a plane taking off from that runway at the same time. The fact that those planes are about to collide is a prediction that we make based on external knowledge and memory that we have, which the “AI” doesn’t, so red flags are raised. AI systems *can* be designed to take this into account, such as by having a separate model trained specifically to watch for runway incursion events, but the CSO describing it as a single AI model that is “omnipresent” demonstrates that they aren’t accounting for any of the known problems, which is mind-bogglingly terrible work that will get people killed. I can easily see interview transcript being read out in a courtroom in a few years after a report digs into the gross negligence that led to hundreds of lives being lost.
The obvious and immediate go to question is - what about loss of electricity? This isn't a paranoid delusion it's a hard fact, sometimes there is no power where and when you need it - especially during disasters, making the disasters far worse, especially in a digital tower. While loss of power for a conventional tower would be horrific - flag signals could be implemented from tower to ground to pilot like the Navy does, but if all you have is ground level and nothing else - good luck.
Genuinely I was looking for this comment and found it 😅 well bro it will help to minimize stress level but what about if and solar flair hit the earth 🌎 or some enemy nation attack electronically like electronic warfare or h crazy hacker I means to say what if someone pull out the power plugs.
If there was an issue at the tower, the same rules would apply as they would today if a tower loses power, radio or radar. The airspace would switch to Class C and pilots would treat it like any uncontrolled airport. The airport would contact the FAA and possibly close the runway and have planes reroute. Everything in aviation assumes things will fail and has contingencies. This is no different than any other tower failure. Honestly, the digital version just gives more options.
@@BlakeHelms I don't know what class C or uncontrolled airport means, but I do know if the "tower" is on the ground or another city, and the power is out far and wide - that's going get ugly fast.
These remote towers may increase efficiency and safety, but occasionally electronic wonders fail. Power is lost, circuit boards short out, cables are severed, and total failure can happen. There are always redundancies built in, but the ultimate backup for air traffic control is the human eye. Heathrow would slow to a crawl without its computers, but it would still be possible to direct an individual plane to a safe landing by human effort if they can see what's going on.
It makes sense when many of the airports (like in northern Norway) only have 1-2 flights pr. day, and even the "busiest" airport only have 7-8 flights in a day.
As a helicopter pilot, this AI system should not ever be replacing humans in the operation loop. We are remarkably good at solving complex problems quickly, and can consider factors that any computer system may not know of. If my aircraft loses it's radio, the air traffic controller can step outside of the tower with a directional light and tell me what to do that way. These computerized systems have had notable success as an integrated component, but should not be used in replacement of controllers. And it'd be nice if they decided to pay controllers more and relax their schedules so they don't have such a high turnover rate. That might be a better solution to the shortage issue than trying to tech-bro some solution.
Is this to save money on the expensive large control towers? If so they can still have controllers at the airport in a cheap room with their computers.
Unfortunately, I think this system will ultimately lead to an increase in air-traffic controller complacency due to: 1) an inunation of data coming in, and 2) an attitude of "Oh, the computer will take control of everything anyway... So I'm just gonna sit here and do nothing." Everything will work great until 5 airliners crash because the computers' A.I. had a glitch. In all reality, there are essentially four main problems in aviation today: 1) Overcrowding of takeoffs and landings 2) Non-standardized terminal and runway configurations 3) Overworked pilots and air-traffic controllers who are forced to pick up the slack due to staffing shortages 4) Haphazard interconnectivity of intermodal transportation in the local area
With complex systems increasing in their complexity and the competency crisis continuing to grow, I'd be concerned about reliability, redundancy, and resiliency. Will their be a backup or failover, not if, but when things go wrong?
To address some questions raised here: What happens during blackout? Ever heard of batteries and generators? What if they get hacked? They are on dedicated connections, NOT connected to the internet. What about rain, snow, ice on the cameras? Wipers and heating. Fog, low visibility? Same as regular tower and humans looking out. Airport shut down, or running at reduced capacity. What if a camera fails? They have multiple overlapping cameras, so that one camera failing will not cripple the system.
I think this is an EXCELLENT system to augment air traffic controllers, especially in airports that are vast and busy. I am sure there are places for this. This might even be a great way to make airports without a ATC on shift be MUSH safer than having pilots land themselves at their own risk. I think this has a place but I do not think that this is a means to replace controllers in seats.
Its possible but the question is if you should. Cameras exist, radar exist, transponder exist. You can make a lot, when you have a lot of control and airtraffic is a highly controlled operation. You can't just put an engine on a few wings and land on an airport. The issue is that a lot of problems will occur. Because the regulating agencies won't trust it, a human still need to do stuff. Usually companies are bad at planning and they will see an operator is not fully used, they look that he will have enough work and suddenly the operators get stressed and they burn out. The tech can fail. With an tower you have windows and a battery operated handheld and you could maintain a limited service. When you have just cameras as windows, you could operate blind but this is very sketchy. The independent handheld also would work a kilometer away. When you have remote towers, this isn't an option anymore. You have 3 points that can fail. The tech at the airport could break down, the remote center can break down or the communication between them. Still an airport without workers is not possible. You need at least need security personell that could arrest someone. You could teach them to take an handheld and announce that the airport is closed. But you went from minimal operation to none. Now imagine a big emergency. Suddenly in a radius, all airports go down and you have a certain percentage of airplanes in that radius that can only land in that radius. With minimal operation, you can get them down, without having a dangoures situation. When everything is remote and you don't have the personell anymore and not enough time to get them to the airport, this is dangoures. These airplanes need to communicate between themselves and sort stuff out. There needs to be someone that takes the lead and organize everything.
Technology will definately help the busiest airport, especially ground control. Visability on a screen will limit operations and flexibility. The use of virtual equipmeny would probably work better but wearing head gear for long periods of time would be uncomfortable. I still hold the future to be a change to aircraft capable of vertical landing and departures instead of long runways. I moved iron birds for 2o years. It will be a long long time before control towers disappear.
I'm doing a PhD in AI. Current machine learning systems are worse at making mistakes, and they are not fully understood. I think it's too early to employ this kind of technology in the airline industry.
@@thiviyanthanapalasingam8683isn't that the point of ML, to mess up as much as possible in a short amount of time in a controlled simulation... it's literally lab testing before going public
@@darealbrianpark i work within an area of machine learning that specialises in safety-critical applications where you need guarantees that machine learning won't fail like healthcare or autonomous driving. The growth of the field has been slow so far because of theoretical challenges making it very difficult to scale to real world applications. It's hard to predict how far we are but I would say at least 10 years. My girlfriend's father works in the airline industry, and he tells me that the airline industry moves very slowly due to safety reasons. So I would say 15.
@@Pfyzeryes, that's generally the principle of software engineering. However, deep neural networks (the type of machine learning models that power many applications that you know of today) are harder to test fully, because it's really hard to understand what they have learned from the data that they have been trained on. In traditional software, this is not the case because a software engineer would write line by line instructions on how to execute something, so testing is much easier. Testing in machine learning is not fully understood. Sometimes a lot of testing is done before deploying it to the public but even then they can fail (e.g. chat-gpt or Google Gemini), and they continuously improve it while they receive feedback from the public. Unlike chatbots, ATC control is more safety critical. Lives can be lost. Hence my worries. No amount of testing can really ensure 100% with deep learning models.
Trusting AI for anything when there are programmers like those of Gemini is like trusting the FAA with Diversity, Equity and Inclusion standards - someone is not going to make it - lots of someones, and then everyone will all say, "this was completely preventable".
There's no room for error in ATC. If one of those remote air traffic systems goes down, it can cause a collision at worst and at best, require pilots to communicate with each other on an agreed frequency instead of ATC. In the case of AI, I could see it really easily making one simple mistake and causing a collision that costs hundreds of lives.
Study the advantages and disadvantages of this technology. One major disadvantage I see is the building has no windows. That and one of these facilities is not at the airport at all. These things likely use huge amounts of electricity. What happens when the electric power fails?
Internet feeds have been throttled and blocked by hackers through node attacks. This system needs a multinode live feed system to be more resistant to such attacks. The other alternative is a direct airport to airport data feed bypassing all internet nodes. Satellite direct data feeds are out of the question due to their extreme costs thought SpaceX Starlink could greatly reduce such coasts.
No thanks. First time a bird craps on a camera, or a heavy rain, or ice or snow cover a lense or a rat chews through a cable. Maybe as a supplement, but for the immediate future, I want the good ol' Eyeballs Mk 1 on the planes.
It probably could be done....not in the US though. In the US, the government contracts are about maximizing profit for the contractor, not delivering a real product. Plus there are other "forces at play" preventing it from happening. Hence the various US attempts have failed.
This sounds like a great idea in terms of helping assist humans with the increase of traffic, and adding extra tools to help increase safety; however, it sounds like an awful idea to completely eliminate the human factor for larger airports like New York and London to name a few. The second any technology fails or allows a disaster to happen all chaos will ensue as AI does not know how to handle such situations.
I still prefer conventional ATC, it’s cheap and simple to maintain. Instead of throwing fund to such advanced ATC I would like to build more runway and expand terminal.
@@machinmon. No, they wont, they cant. What is an AI powered aircraft gonna do during a hijacking? Medical emergency or pregnancy? AI would have crashed 1549 trying to get back to the airport if it weren't for Sully knowing to land in the river.
With the implementation of ADS-B on all flights, now there is an army of people capturing radar data with USB SDR devices and we have a near world wide system monitoring every plane movement, to the exact location and starting to get all airport vehicle info too. I am one of thee radar feeders, and I feed to 5 websites where anyone can go online and see any info àbout a flight and even see the actual picture of the plane thanks to plane spotters that feed pics to online databases. Then back that to the FAA databases info and now I see a plane ahead of my house, I click on it, sée the picture, owner, speed, alt, to/from, the entire history of that plane, where it landed when and stuff. Also see any failures of certificqtes for it
I think it's reasonable to have concerns and be very cautious about any new technology, which should be introduced carefully. But outright rejection (per many of the comments) is, imho, naive. If we rejected all new technology which has a risk to human life (such as this one) we would not have auto-land (or even autopilot), RADAR, ILS, GPS navigation, fly-by-wire, pressurised cabins, just to name a few. And for those saying "what happens when there is an outage", that is already a potentially fatal situation even today. There will be issues, possibly bad ones,, and each time that happens the overall system improves. The pilot who does the MentourPilot TH-cam channel is excellent as describing how this happens.
I get that digital towers can reduce the cost on building new infrastructure but it increases the margin error and make airports vulnerable to Cyber attacks and terrorism. Nice idea, but they should think on the possible risk of implementing this system as is. On the other hand, I guess it will be more beneficial having the towers with the cameras and humans monitoring closely. 4:34
Places like that would, of course, have both battery and generator backup. This is proven technology, in use in places like hospitals, banks, tv-stations, radio stations and many more all around the world. The battery provides power until the generator has started up. Nothing new about that. Even some of my friends (living in areas prone to power outages) have UPS-units connected to their computers so that they can safely save their work and shut down the computer.
To be honest, I am missing the "AI" part here. All I see is image recognition, object detection, Human detection etc. What exactly is the contribution of AI here?
It sounds like a good idea *UNTIL* some state actor (or not!) hacker gets into the system and causes no end of trouble. Remember _Die Hard 2_ ? While that scenario is obviously complete Hollywood fiction, with this system, it may not be so far-fetched after all.
@@redbean9410 Well, the "human eyeball" factor to override the computer hacking. That's why I still foresee airports still built with control towers for the foreseeable future.
In the case of unanticipated external consequences, force majeure, etc., analog technology still holds value. The over reliance on technology in any industry could have catastrophic consequences when the technology goes down. We cannot take away the human factor as part of the overall equation. At least not yet.
I'm all for digital air towers and improvements, but decentralizing air traffic control into one location instead of local air traffic controls at each airport seems like a great way for one internet connection to ruin 20 airports
Redundancy would address this concern. Also, I'm not convinced that an internet outage wouldn't affect airports today.
London City has a triple connection. 2 independent optical links that run separate routes and if both fail, a wireless link.
And you don't need many controllers for airports that only see a few aircraft a day.
Another job that will be replaced by AI. :/
@@Hans-gb4mvwhat if all three fail
@@LeeeroyJenkinsit’s not really replacing people. You still need people in the inner workings and stand by. People are still there doing the job so to speak. It’s just that the use of AI streamlines and automates the process to make it more efficient and possibly safer.
As an airline pilot I can say that for sure at least in the US, the ATC system is under extreme pressure. In 20+ years of flying I've never heard the frequencies be so congested and the controllers sounding so winded at peak hours, and they often have to resort to many delay vectors even on perfectly clear days. It's not their fault, just the nature of the beast. The FAA innovates at the speed of molasses due to all the regulatory red tape involved which is great for safety but terrible for efficiency. In the early 2000s they came out with the new "nextgen airspace" moniker and promised very lofty things to be in place by 2020. It's now 4 years past that and we've hardly seen any of that come to fruition, aside from ADS-B which while great has also has placed quite the burden on many other areas like GA pilots, drones and RC hobbyists, all of which I'm also involved in. Some smaller, currently untowered airports would certainly benefit from this technology, but it's still far far away from being viable for any of the top 100 busiest airports in the world. I've dabbled plenty with chatgpt and Gemini and as impressive as they can be they're extremely stubborn and rigid at times, and seem completely unable to improvise in some complex situations. This could be a life and death thing in an emergency when airborne.
As a AI scientist, I agree with you. A lot of the machine learning technology are not mature enough. The way they are trained makes it impossible to act on their own in new situations that have been not been encountered by these models. I wouldn't feel safe if they deployed these models to safety critical applications like ATC. It's way too early.
you know what they say pal, safety regulations are written in blood. Better slow than costly I guess
I believe they're purposely creating a pressured environment for this very reason.
@@Hans.Dewitt As disturbing as it sounds, I very much do agree.
No existing ATC has anything to worry about during their careers. But I remember having a conversation with a tower controller once (I’m a center controller) and somehow we ended up talking about why I wouldn’t recommend anyone new to go to the tower route. Among many reasons, I also said how the towers are going to be the first to go. It will start with the small towers first until only the Core 30 towers are left.
A digital tower should be an aid/supplement, not a total replacement of an actual control tower. (Especially on Major Airports)
an air traffic control tower is often the icon of an airport anyway, i think for major airports the digital tower would probably be a supplement to real towers
i only see purely digital towers being in use in remote airports, where they get maybe half a dozen flights a day, because the number of flights are so low, a tower out of order wont be detrimental to safety
and... if all those camera's fail or get hacked.. Nobody could see the planes. 😮
And why is that? We are not an expert they dont care about our opinions.
That’s exactly what will happen. This is pathetic click bait you fell for
@@qwerty-yv8wmI am an expert, you might be just a mc Donald’s employee and that’s fine, they don’t care what you say you’re right.
Not being in the tower and able to look with your own eyes seems like asking for a digital disaster to happen with no immediate work-around.
Cyber attack, power-outage and the backup fails at some point… someone’s gonna be sent running up that decommissioned tower and they’re not going to be in shape to do it.
They'll also have to keep airport level security in the remote towers to be safe.
Power outage can incapacitate traditional towers as well
I think that's the point. You wont need workers, but you will need a few specialized A.I. engineers. To supervise .....A.I.
At the end of the video, they showed an ATC moving a window of one portion of the airport to another section of the screen, essentially moving a taxiway virtually. I think this type of thing can only lead to confusion and deadly accidents as the ATC lose track of the relative relationship between aircraft and airport locations. Seems like the whole point of those giant screens is to display the entire airfield at once and maintain situational awareness. The option to transpose taxiways is dangerous.
That's what people said about autopilot.
4:53 I'm a pilot in Colorado and flying into Northern Colorado (FNL) was always interesting. It's treated like a Class D airport while not technically towered. ATC is on-site in trailers with a camera pole and they'll request you report X miles out since they don't have radar.
so would you say your job is like driving but in the sky
Im sure AI assisted with human controllers would be okay. Like an extra set of digital eyes cant hurt. But still need humans as backups. Tech is prone to mistakes, crashes, errors, that sometimes arent fixed immediately like an airport would need. Id say its a fine line between mixing the two
So much tech is already essential. I wonder If one day much of it will go down from solar flares or something.
hmm, I think the AI is the backup for the primary human decision maker...
If so makes an error that resulted in fatalities then who is held accountable?
@@electrology Accountable is an adversarial word. We learned long ago that flight incidents are caused by a chain of events. The entire chain must be discovered and analyzed. Not to assign accountability, but to take measures to prevent the chain of events from happening again.
@@electrology if it was only AI running the show most likely be considered an industrial accident.
The first digital tower was in Sweden where we have had them for years. They work very well for our airports in the north.
That's because it's Sweden and no one cares enough to mess with anything.
no wonder why sweden has a massive immigration crisis
I can see virtual towers on site where a normal tower sits, but it is madness to have virtual towers miles away... network connectivity is far from a sure thing, even when multiple redundancies are in place.
There are secure networks with guaranteed level of service.
You might could read up on how NORAD communicates with America's armed forces clear across the world.
I don't think America will introduce this concept unless it's better in many many ways.
@@DinoAlberiniIn the IT and telecom world, "guaranteed level of service" means that if they don't meet that level of service, they discount the fee they're charging you. It doesn't mean it's impossible to have an outage.
@@macsound and that’s why you have redundancies
@@DinoAlberini Redundancy quite literally means relying on multiple systems to reduce the impact of failure. If a centralised ATC service fails, the entire country loses the capability to direct traffic into and out of airports. If you have a team at each airport, failure means just one airport goes offline. And nothing is quite as robust as being able to look out the window.
Makes sense. I know there's a lot of dependence upon electronics for aviation. But even still, what happens when the power is cutoff? Being able to see it yourself seems like an excellent backup plan, and would require someone be there to see what's going on.
So call centers for airplanes. How long before that's outsourced overseas?
I love that everything is becoming so digital but I feel airports are now going to get so much bigger without the towers
As a 16 year veteran controller in the USA I can admit our system is saturated and frequencies are packed in the busier air spaces.
And I hate to think that I can not be replaced but at the end of the day, the pavement at the airports can only handle so many planes within a limited time.
We get so efficient and introduce more aircraft and sort of get back to the original delay.
I’m curious to see where this goes!
AI is rapidly improving.
I think AI is going to rock your world the way computers improved productivity when they were introduced. Safety first.
The railroads use to mandate a caboose with a live operator in them in early trains. Then they went without operator because of technology then they did away with caboose. LOL
@@28704joe I think so too. The timeline is the question. Sooner than I think I’m sure.
THE SAME REASON THIS IS A BAD IDEA is the same reason it is a BAD IDEA to paint all your car windows with black paint, attach cameras all over the car, then installing monitors in the car, and then taking the car out for a drive near a school zone when kids are getting out of school. If you find that scary, relate that same scenario to airplanes.
I did this, and it is my opinion that the children had an absolutely wonderful time!
Whatever helps with the delays
and safety *
I rather have delays than safety.... I can live with delays but we won't live without safety
fewer flights
The ignorance
Fewer flights, more trains, and stop using airports on the coast as major hubs.
I dont think air tower control will disappear. You always need someway to do things manually. If those system are in maintenance or fail. You need something to do by the eye.
What happens when a bird lands near the tower, or bad weather blocks the camera? What about when the power goes out? I can't be the only one thinking about this.
Yeah, in German we say "verschlimmbessern" wich roughly translates to improving something, while making it worse in the process.
If bad weather blocks the cameras, it would also block the tower from seeing too and if the power goes out at the airport then so does the towers power at the airport
But if you have multiple Digital control centres and one goes down, the others can take over their airport operations
You're not. This is tech companies trying to sell technology that doesn't really work to unsuspecting officials. I've seen this countless times in manufacturing.
Automatically guided carts that 1) couldn't avoid people 2) couldn't emergency stop 3) worked way below the projected line speed.
"What happens when a bird lands near the tower"
I'd assume multiple cameras in different position with overlapping fields of view would solve that.
"bad weather blocks the camera?"
Bad weather (heavy fog, rain or snow) will also blind the controllers looking from the ATC tower. I work at an airport and there's a reason why LVP (Low Visibility Protocol) is a thing and every emplyee knows what that means: different rules for going about your work.
"What about when the power goes out?"
Airports all ready have their own backup power generators for just that eventuality. And if proper safety and control systems can't be brought back up quickly enough the planes will divert to a backup airport.
"I can't be the only one thinking about this."
No you are not and before anythng like this is actually adobted it will go trough extrimely rutheless testing. There are international and national authorities that will need to accept this kind of tech before it'll be accepted.
Low visibility already causes accidents on airports even now. Non-visual sensors can ease this issue by providing alternative means of information. This is a plus. And if power outage means no screen is working, it also means no radio is working. So no one would be dare to install any vital electronics without an alternative power source. These are not problems unique to smart technology.
What happens when there is an outage? Or perhaps a cyber security breach?
It would be much safer to use line of sight from a tower and standard radio communications to control the traffic. This seems ripe for failure. Sometimes things should stay the same. Not to say we can’t empower controllers to do an amazing job by empowering them with technology but removing them from the airport and staging 70 miles away is just asking for problems.
No problems in the three years it's been operating.
@@colinpotter7764 this is the same arguement as “nOtHiNg BaD hApPeN-d sO iT wOnT”
@@master1387 no it isn't. Let's ban all aircraft because they may crash shall we?
What happens when it’s foggy? Or if someone comes to work drunk? Humans are just as fallible than computer systems. It’s about costs and capabilities and this video makes it clear there’s humans manning ATC functionality, they just aren’t using binoculars, squinting, and coms to figure out what’s going on…
@@bobthemagicmoose the only problem when it's foggy is the pilot may not get to work. If he's drunk then there are bigger problems.
Hey! One of my local airports was on the list at 4:44! I feel seen
Could you see my house from there?
Nobody thought the array of screens with large gaps between them was a window come on writers!
Yup!
Speak for yourself
I have never seen a 360° window.
So the "gaps" could be called window frames.
What they have done in UK and elsewhere is designed a system where controllers can be remote workers. Can the controller be given computers and displays so they can work from home in their pajamas?
Let's go a step further. It will outsource the at job to India or other cheaper countries
Are you writing this in your pajamas? Now be honest. :)
Great way to protect your citizens and national infrastructure. @@imfunatparties9463
Fascinating video.
It’ll be interesting to see how quantum radar (which is 65,000 more sensitive than regular radar) will be integrated into Air Traffic Control (ATC) systems …
Give a whole new meaning to work from home
i think that this is great but... if any power outage happens you can't see the planes... so, this should be an aid NOT a replacement. Eventually the tower will be a watch area but the main comms can happen in the digital room
Yeah… a remote station works great until some construction worker accidentally smucks a fiber line when doing some ditch maintenance. Hope they have numerous connections ready to go for backups.
Not sure if this new digital tower technology is very comforting,. As a passenger, I see so many things that can go wrong.
Same was said of autopilot.
@@28704joe yes, but they didnt remove the pilots and let autopilot do the whole job did they
I think it’s funny they use display tablets designed for artists/designers to move around controls😅
Especially if they can get multi-spectral cameras (infrared, etc.) in on the act, they'll have better visibility, WRT what is where, during adverse weather conditions. The would enhance safety, specially in nighttime, fog, etc.
The next thing they need to add, and the NextGen system in the USA is supposed to provide, is digital text information between ATC computers and pilots. A pilot could notify the ATC system that, for various reasons, they can't make their takeoff slot. That open slot would be notified to all other aircraft, such that another aircraft could take that slot (such as the next in the queue, assuming they'll be ready in time) and things can move along more smoothly. Meanwhile, as various aircraft move up, a slot would open later in the queue, such that the earlier aircraft can still get one in a timely fashion. This could all be handled by computers, without needing human ATC operators getting involved. Scheduling the various flight out, queuing planes onto the taxiways, etc. could all be handled to greater degree by computers and the pilots, reducing the ATC operators' workloads.
So long as digital networking tech is performant (enough bandwidth and sufficiently-low latency) and reliable, it should be possible to put the actual ATC operators just about anywhere. It might become possible to have a handful of ATC centers in an entire country (as demo'ed in Norway) coordinating all the major airports. But there would need to be redundant networking; working for internet service providers over the years, I've been amazed how many times a "cable-seeking backhoe" managed to find the actual cable providing the company's internet backbone and take them completely off the net. Having a pair of different backbone providers, through different cables, coming in through different physical routes, was the only way to avoid that.
You’re describing the ATFM, air traffic flow management. As far as infrared, that’s ubercool and I’m totally for it but radar and ADS-B are all they’ve given us.
Given all the things happening in the Cyber world currently, I would say we still need a tower. This could be a good add-on but not an alternative.
It is only a matter of time that it will be the alternative. There isn't a reason why it couldn't replace people.
When the NATs system works it’s all well and good, but when it fails it’s a right mess, as it did on August 28th 2023. The incident was triggered by the inability of the Nats computer system to process unusual, but correct, data in a flight plan submitted for a plane crossing UK airspace to Paris from Los Angeles. Two waypoints with an identical code triggered a “critical exception error” that caused the primary and secondary systems to disconnect themselves and stop working. On-site technicians couldn’t resolve the issue, and the specialist engineers were working from home. It took 90 mins for an engineer to arrive on-site, and it took a further four and a half hours till the fault was resolved and UK airspace could operate at normal capacity. But the damage was done, due to safety regulations restricting staff flying hours, planes being in the wrong place, and other subsequent logistical issues, over 3000 flights in/out of the country were cancelled, 700 000 passenger’s stranded and disruption for a 3 days. A real utter calamity, and that’s the risk you run with centralised operations, yes they are more efficient, cheaper to run, but when things go South, the negative effects are more far reaching and impactful.
Am I not looking in the right place? I don't see links to sources. I want to see the FAA outdated tower list (without having to search for it myself).
(FYI - www.transportation.gov/sites/dot.gov/files/2022-03/FAA_Buget_Estimates_FY23.pdf)
There is so much the FAA could do
Let’s start with PDC, D- ATIS how hard can that be
I can see it at smaller GA airports with high traffic
But how do you keep cameras operational unless you have lots of redundancy
Very interesting. It would also be interesting to see how they are overcoming managing flights in real time when a network outage happens. Redundancy?
It’s a great idea as long as everything works perfectly. If you have say a contractor that wipes out the fiber optic cable to the master control facility then there might be some problems. If there are problems with the radar or ADSB transponder then there might be problems. If the local radar or ground surveillance radar fail there might be problems.
I wonder what kind of safety nets are in place.
For better visibility, identification, and logistics - great idea! But I'd never fly again if actual air or traffic ground coordination was controlled by AI or any computer. A computer can identify hazards and alert the ATC, but shouldn't make those decisions on its own.
Also: want to cause some trouble? Hack one of those remote tower camera feeds and watch the economy crumble
So with apple vision pro, they can work from home
How cool and nice this looks all the AI and digital distance control. this AI/Digital tower operation is a very big risk and target too for hackers and digital warefare, they better invest good in digital security on all layers of use and training for people how to work the system safe and secure !
But what happens if blackout occurs?
lol then a server goes down and chaos and death ensues… and the airport will just say “oops”
why do you think they wouldn't consider this into account?
some of the best engineers in the world are working on this, don't you think they'd think of that and have redundancies? your life for the past 20 years has been run via servers, how often do you see them go down?
The ATC is relying on servers today as well. Aircraft even more so. They rarely “go down”.
@@DinoAlberini "Rarely" isn't never. The human eye is the ultimate ATC backup.
@@Bobrogers99 it is. When the tower is close enough. On major airports you rely on technology to monitor traffic.
How does NATCA feel about this?
And what if the cameras fail?
So what happens during a power outage? internet outage? or damaged camera/ computer?
how will airplanes get to the ground if the only have 15 mins and all of those happened all at once?
about time!
I think this move needs some strong pros and cons analysis. Yes shifting to full-automation will make analysis and planning more accurate and quick, but we'd be over relying on 2 things: electricity and IT. Once either of the two has a problem, there's almost nothing anyone can do in case of an emergency..?
Why they don’t put little feet on the towers, so they move anywhere?
What happens when the camera fails? Will you have to go outside?
When the power goes out, system gets hacked or goes down... CHAOS! very very risky.. What are the safeguards?
Don't worry, a wizard will solve that problem.
Power shortage or things like that can easily resolve
Agreed, IT issues have grounded flights in the past. We should always have backup systems for IT failures. Computer vision isn't 100% reliable either. I'm guessing this "digital tower" thing is a push to cut costs on tower construction and to outsource air traffic control jobs.
2:38 Hearing the CSO of the system refer to the AI as “monitoring a wide range of things simultaneously, whereas a human has to focus on particular areas” and “omnipresent” is extremely worrying! It shows both a near-total lack of understanding of how it works and a massive amount of complacency (ironically, a lack of focus on the danger). Those are two of the main ingredients for air industry disasters.
The entire basis for the tech they call “AI” in these systems is that it focuses similar to a human, and lacks all forms of short and long-term memory (the “keep in mind” context of a situation).
For example, an AI model will “happily” place a callsign tag on one plane crossing a runway and another tag on a plane taking off from that runway at the same time. The fact that those planes are about to collide is a prediction that we make based on external knowledge and memory that we have, which the “AI” doesn’t, so red flags are raised.
AI systems *can* be designed to take this into account, such as by having a separate model trained specifically to watch for runway incursion events, but the CSO describing it as a single AI model that is “omnipresent” demonstrates that they aren’t accounting for any of the known problems, which is mind-bogglingly terrible work that will get people killed.
I can easily see interview transcript being read out in a courtroom in a few years after a report digs into the gross negligence that led to hundreds of lives being lost.
Marvelous 😮.Unlike self driving cars ATC is definitely an area AI could extensively be used.
The obvious and immediate go to question is - what about loss of electricity? This isn't a paranoid delusion it's a hard fact, sometimes there is no power where and when you need it - especially during disasters, making the disasters far worse, especially in a digital tower. While loss of power for a conventional tower would be horrific - flag signals could be implemented from tower to ground to pilot like the Navy does, but if all you have is ground level and nothing else - good luck.
Genuinely I was looking for this comment and found it 😅 well bro it will help to minimize stress level but what about if and solar flair hit the earth 🌎 or some enemy nation attack electronically like electronic warfare or h crazy hacker I means to say what if someone pull out the power plugs.
@@harshraj8729 Blank account bot. Sigh. Why are bots always programmed to be negative never make sense? AI is stupid not intelligent.
If there was an issue at the tower, the same rules would apply as they would today if a tower loses power, radio or radar. The airspace would switch to Class C and pilots would treat it like any uncontrolled airport. The airport would contact the FAA and possibly close the runway and have planes reroute. Everything in aviation assumes things will fail and has contingencies. This is no different than any other tower failure. Honestly, the digital version just gives more options.
@@BlakeHelms I don't know what class C or uncontrolled airport means, but I do know if the "tower" is on the ground or another city, and the power is out far and wide - that's going get ugly fast.
you think that the engineers who have years of experience and other people's entire lives worth of knowledge didn't take this into account?
You still need a tower to put the camera on.
These remote towers may increase efficiency and safety, but occasionally electronic wonders fail. Power is lost, circuit boards short out, cables are severed, and total failure can happen. There are always redundancies built in, but the ultimate backup for air traffic control is the human eye. Heathrow would slow to a crawl without its computers, but it would still be possible to direct an individual plane to a safe landing by human effort if they can see what's going on.
Oh yeah, that sounds real safe cutting costs so the controllers can control aircraft from three airports at a time super smart
It makes sense when many of the airports (like in northern Norway) only have 1-2 flights pr. day, and even the "busiest" airport only have 7-8 flights in a day.
As a helicopter pilot, this AI system should not ever be replacing humans in the operation loop. We are remarkably good at solving complex problems quickly, and can consider factors that any computer system may not know of. If my aircraft loses it's radio, the air traffic controller can step outside of the tower with a directional light and tell me what to do that way. These computerized systems have had notable success as an integrated component, but should not be used in replacement of controllers. And it'd be nice if they decided to pay controllers more and relax their schedules so they don't have such a high turnover rate. That might be a better solution to the shortage issue than trying to tech-bro some solution.
Is this to save money on the expensive large control towers? If so they can still have controllers at the airport in a cheap room with their computers.
Automation is all good in theory ! but what if all camera goes down/destroyed/malfunction/hacked? There should always be a manual backup plan.
Unfortunately, I think this system will ultimately lead to an increase in air-traffic controller complacency due to: 1) an inunation of data coming in, and 2) an attitude of "Oh, the computer will take control of everything anyway... So I'm just gonna sit here and do nothing."
Everything will work great until 5 airliners crash because the computers' A.I. had a glitch.
In all reality, there are essentially four main problems in aviation today:
1) Overcrowding of takeoffs and landings
2) Non-standardized terminal and runway configurations
3) Overworked pilots and air-traffic controllers who are forced to pick up the slack due to staffing shortages
4) Haphazard interconnectivity of intermodal transportation in the local area
Yea very safe. What happens when the lights go out or a monitor dies?
Is everyone going to run upstairs to the analog tower.
I wonder what could possibly of happened when the line dropped on the graph
isn't a tower needed to mount the cameras?
With complex systems increasing in their complexity and the competency crisis continuing to grow, I'd be concerned about reliability, redundancy, and resiliency. Will their be a backup or failover, not if, but when things go wrong?
To address some questions raised here:
What happens during blackout? Ever heard of batteries and generators?
What if they get hacked? They are on dedicated connections, NOT connected to the internet.
What about rain, snow, ice on the cameras? Wipers and heating.
Fog, low visibility? Same as regular tower and humans looking out. Airport shut down, or running at reduced capacity.
What if a camera fails? They have multiple overlapping cameras, so that one camera failing will not cripple the system.
Tech will never make it obsolete
Said the eight track to the MP3
I’m surprised the WSJ didn’t cover the fact that RTX is standing up a new test remote tower at ACY, in partnership with Frequentis from Austria
And what happens if you lose camera feed to one or more cameras, then what ?
I think this is an EXCELLENT system to augment air traffic controllers, especially in airports that are vast and busy. I am sure there are places for this. This might even be a great way to make airports without a ATC on shift be MUSH safer than having pilots land themselves at their own risk. I think this has a place but I do not think that this is a means to replace controllers in seats.
Its possible but the question is if you should. Cameras exist, radar exist, transponder exist. You can make a lot, when you have a lot of control and airtraffic is a highly controlled operation. You can't just put an engine on a few wings and land on an airport.
The issue is that a lot of problems will occur. Because the regulating agencies won't trust it, a human still need to do stuff. Usually companies are bad at planning and they will see an operator is not fully used, they look that he will have enough work and suddenly the operators get stressed and they burn out.
The tech can fail. With an tower you have windows and a battery operated handheld and you could maintain a limited service. When you have just cameras as windows, you could operate blind but this is very sketchy. The independent handheld also would work a kilometer away. When you have remote towers, this isn't an option anymore. You have 3 points that can fail. The tech at the airport could break down, the remote center can break down or the communication between them.
Still an airport without workers is not possible. You need at least need security personell that could arrest someone. You could teach them to take an handheld and announce that the airport is closed. But you went from minimal operation to none.
Now imagine a big emergency. Suddenly in a radius, all airports go down and you have a certain percentage of airplanes in that radius that can only land in that radius. With minimal operation, you can get them down, without having a dangoures situation. When everything is remote and you don't have the personell anymore and not enough time to get them to the airport, this is dangoures. These airplanes need to communicate between themselves and sort stuff out. There needs to be someone that takes the lead and organize everything.
Technology will definately help the busiest airport, especially ground control. Visability on a screen will limit operations and flexibility. The use of virtual equipmeny would probably work better but wearing head gear for long periods of time would be uncomfortable. I still hold the future to be a change to aircraft capable of vertical landing and departures instead of long runways. I moved iron birds for 2o years. It will be a long long time before control towers disappear.
Best of luck US! :) We'll let you know how a few decades of this is working. Love, Norway
Informative thanks
the netherlands has a remote tower for one of the airstrips
Agreed. Humans in the ATC tower make so much mistakes because they get tired. America should gradually develop the system as the technology develops.
I'm doing a PhD in AI. Current machine learning systems are worse at making mistakes, and they are not fully understood. I think it's too early to employ this kind of technology in the airline industry.
@@thiviyanthanapalasingam8683 thanks for sharing. How long do you think will it take for ATC tower to be automated?
@@thiviyanthanapalasingam8683isn't that the point of ML, to mess up as much as possible in a short amount of time in a controlled simulation... it's literally lab testing before going public
@@darealbrianpark i work within an area of machine learning that specialises in safety-critical applications where you need guarantees that machine learning won't fail like healthcare or autonomous driving. The growth of the field has been slow so far because of theoretical challenges making it very difficult to scale to real world applications. It's hard to predict how far we are but I would say at least 10 years. My girlfriend's father works in the airline industry, and he tells me that the airline industry moves very slowly due to safety reasons. So I would say 15.
@@Pfyzeryes, that's generally the principle of software engineering. However, deep neural networks (the type of machine learning models that power many applications that you know of today) are harder to test fully, because it's really hard to understand what they have learned from the data that they have been trained on. In traditional software, this is not the case because a software engineer would write line by line instructions on how to execute something, so testing is much easier. Testing in machine learning is not fully understood. Sometimes a lot of testing is done before deploying it to the public but even then they can fail (e.g. chat-gpt or Google Gemini), and they continuously improve it while they receive feedback from the public. Unlike chatbots, ATC control is more safety critical. Lives can be lost. Hence my worries. No amount of testing can really ensure 100% with deep learning models.
Trusting AI for anything when there are programmers like those of Gemini is like trusting the FAA with Diversity, Equity and Inclusion standards - someone is not going to make it - lots of someones, and then everyone will all say, "this was completely preventable".
There's no room for error in ATC. If one of those remote air traffic systems goes down, it can cause a collision at worst and at best, require pilots to communicate with each other on an agreed frequency instead of ATC.
In the case of AI, I could see it really easily making one simple mistake and causing a collision that costs hundreds of lives.
Study the advantages and disadvantages of this technology. One major disadvantage I see is the building has no windows. That and one of these facilities is not at the airport at all. These things likely use huge amounts of electricity. What happens when the electric power fails?
Good to see unions not interfering with progress.
As an airline pilot The US ATC is already been antiquated
Internet feeds have been throttled and blocked by hackers through node attacks. This system needs a multinode live feed system to be more resistant to such attacks. The other alternative is a direct airport to airport data feed bypassing all internet nodes. Satellite direct data feeds are out of the question due to their extreme costs thought SpaceX Starlink could greatly reduce such coasts.
Where would the cameras go if you have no tower!
Good for a backup. Computers, cameras, and electronic systems always go down, always.
No thanks. First time a bird craps on a camera, or a heavy rain, or ice or snow cover a lense or a rat chews through a cable. Maybe as a supplement, but for the immediate future, I want the good ol' Eyeballs Mk 1 on the planes.
It probably could be done....not in the US though. In the US, the government contracts are about maximizing profit for the contractor, not delivering a real product. Plus there are other "forces at play" preventing it from happening. Hence the various US attempts have failed.
What you really need is this system so they don't LOSE your luggage; NOW THAT would be an improvement.
This sounds like a great idea in terms of helping assist humans with the increase of traffic, and adding extra tools to help increase safety; however, it sounds like an awful idea to completely eliminate the human factor for larger airports like New York and London to name a few. The second any technology fails or allows a disaster to happen all chaos will ensue as AI does not know how to handle such situations.
But it got the Radar for tracking the Aircraft in Airspace to smoothen the Flight ✈ Paths, takeoff and landing in all weather environment.
I still prefer conventional ATC, it’s cheap and simple to maintain. Instead of throwing fund to such advanced ATC I would like to build more runway and expand terminal.
Don't worry, soon all planes will be AI anyway, don't need remote interfaces
@@machinmon. No, they wont, they cant. What is an AI powered aircraft gonna do during a hijacking? Medical emergency or pregnancy? AI would have crashed 1549 trying to get back to the airport if it weren't for Sully knowing to land in the river.
With the implementation of ADS-B on all flights, now there is an army of people capturing radar data with USB SDR devices and we have a near world wide system monitoring every plane movement, to the exact location and starting to get all airport vehicle info too. I am one of thee radar feeders, and I feed to 5 websites where anyone can go online and see any info àbout a flight and even see the actual picture of the plane thanks to plane spotters that feed pics to online databases. Then back that to the FAA databases info and now I see a plane ahead of my house, I click on it, sée the picture, owner, speed, alt, to/from, the entire history of that plane, where it landed when and stuff. Also see any failures of certificqtes for it
I think it's reasonable to have concerns and be very cautious about any new technology, which should be introduced carefully. But outright rejection (per many of the comments) is, imho, naive. If we rejected all new technology which has a risk to human life (such as this one) we would not have auto-land (or even autopilot), RADAR, ILS, GPS navigation, fly-by-wire, pressurised cabins, just to name a few. And for those saying "what happens when there is an outage", that is already a potentially fatal situation even today.
There will be issues, possibly bad ones,, and each time that happens the overall system improves. The pilot who does the MentourPilot TH-cam channel is excellent as describing how this happens.
I get that digital towers can reduce the cost on building new infrastructure but it increases the margin error and make airports vulnerable to Cyber attacks and terrorism. Nice idea, but they should think on the possible risk of implementing this system as is. On the other hand, I guess it will be more beneficial having the towers with the cameras and humans monitoring closely. 4:34
Sounds like a fantastic opportunity... for hackers and bad actors. Aviation has been a gold standard of safety and reliability. Proceed carefully!
Let's build a tower to replace a tower! Great idea, till the power goes out and planes need to land.
Places like that would, of course, have both battery and generator backup.
This is proven technology, in use in places like hospitals, banks, tv-stations, radio stations and many more all around the world.
The battery provides power until the generator has started up. Nothing new about that.
Even some of my friends (living in areas prone to power outages) have UPS-units connected to their computers so that they can safely save their work and shut down the computer.
To be honest, I am missing the "AI" part here. All I see is image recognition, object detection, Human detection etc. What exactly is the contribution of AI here?
The new Western Sydney International Airport will have a fully digital and remote control tower
I mean I'm an infosec dude so like I'm not against new tech in general but this seems like a way for a small internet outage to cause huge problems.
It sounds like a good idea *UNTIL* some state actor (or not!) hacker gets into the system and causes no end of trouble. Remember _Die Hard 2_ ? While that scenario is obviously complete Hollywood fiction, with this system, it may not be so far-fetched after all.
what's stopping a hacker right now from taking over current towers?
@@redbean9410 Well, the "human eyeball" factor to override the computer hacking. That's why I still foresee airports still built with control towers for the foreseeable future.
@@Sacto1654 human eyeball would be pretty useless without communications. I don’t disagree with you, but I’m sure many redundancies would be in place
@@Sacto1654it’s a closed system. If you want to hack it, you’d need to be physically there.
All fun and games until the camera feeds get disrupted.
In the case of unanticipated external consequences, force majeure, etc., analog technology still holds value. The over reliance on technology in any industry could have catastrophic consequences when the technology goes down. We cannot take away the human factor as part of the overall equation. At least not yet.
Skynet is a reminder to never ever completely take away the human element from controlling critical parts of the national infrastructure or security
I wonder what would happen, or the lack of if someone sets off an EMP at that airport. Just how blind is blind for the systems placed there?