it is crazy to me that all of this still relies on humans giving verbal clearance over radio.
Shouldn’t there be a universal standard to communicate clearances to the computer systems of the airplanes and other runway vehicles? And shouldn’t there be a small, well-tested core system that makes sure clearance never conflicts?
How on earth does all this still depend on skilled individuals not making a mistake?
> Shouldn’t there be a universal standard to communicate clearances to the computer systems of the airplanes and other runway vehicles? And shouldn’t there be a small, well-tested core system that makes sure clearance never conflicts?
ATC entering instructions into a computer, then pilots reading and interpreting from their computer, and then operating their plane, probably is slower and the context switches between the computer screen and what's outside the window may lead to errors.
There are over 100,000 commercial flights each day (worldwide):
I didn’t say that anyone should enter anything manually into a computer, nor did I claim that a human reading a computer screen is how an automated runway system would ensure safety.
As an aside, humans are not doing a good job at driving cars and should leave that to the computers.
Where I drive daily I constantly have to make decisions which no self-driving car system I know about can possibly do. They don't foresee upcoming gridlocks caused by you driving completely legally but blocking (again, legally) a car from crossing to a side road, that car again blocks the car behind it, which blocks the car in front of you from doing a left turn, which means that you can't move and unlock the gridlock. That's just one on my commute. Then there's driving uphill on snow, seeing a person walking on the other side of the road and a car coming in the other direction. You foresee that this car will have some problems braking for that person (which would be the normal procedure - I'm coming the other way, that car shouldn't try to pass the person). So, what I do is to slow down and give that other driver confidence to pass. And sometimes I have to do the other thing - speeding up a bit in order to create room for another car to let that car avoid problems (if I were a computer I would have no reason to think about that).
This is also something happening all the time where I drive. And tons of other understanding-the-traffic cases.
Self-driving systems only follow the rules. That is not sufficient. When general artificial intelligence eventually arrives - that's when we get self-driving cars better than humans, in any other cases than for trivial cases. And yes, computers could drive cars in trivial places, it's just that those places are pretty much non-existing where I live and drive.
While everything you mention are signs of good, defensive driving and seeing more than "Just following the rules," average human drivers don't do what you do. And that's what the automated systems are compared to.
So again, while I think ultimately the automated systems should also do what you're talking about, right now, if the other article is true in its claims that Waymo is safer, it seems sufficient to follow the rules to achieve a certain level of safety.
[other drivers] But they do - where I live. It's very rare to see people not doing that, so much so that everyone really notices.
Edit: And the point is really that strictly just following rules (i.e. just drive if you're in the right and the signs say so) will cause accidents and worse. For others, not necessarily myself. On my commute and nearby.
There's nothing so far indicating that they have, at this point. Though I doubt predictive algorithms would work - there are all kinds of situations, all slightly different, and lots of them about understanding what someone else might be thinking. As I mentioned, I think we'll need general artificial intelligence to do this properly.
No, never been to SF, I have no idea how people drive there or what the conditions are. Except that they're very different from where I drive (except when I'm in the town centre) - roads without sidewalks, mostly no signal lights, snow a large part of the year, and other special conditions.
That is not quite accurate. There are certainly standards for airport markings, but there are many non-standard airports... especially when flying at anything smaller than an international airport.
(And yes, even airliners will sometimes land at small regional airports.)
Is that data controlled for environment? I.e. it must only compare Waymo driving with human drivers in exactly the same environment. Same road, same conditions, same time, etc. Waymo doesn't take any difficult routes.
Not to mention the fact that when it encounters a situation never seen before it just "freezes" - happened a few times and caused traffic jams.
What I give these cars is their ability to react "instantaneously" to emergency situations like automated emergency braking when there is an obstacle in the front of the car or the car in front suddenly brakes.
IMHO, there was not even enough data to draw this conclusion from the study. 3.8 million miles looks like a lot, but it is laughable for traffic safety conclusions.
And of course Waymo is going to milk as much as possible from their study, we just shouldn't trust it blindly.
We need far more data to say this with certainty. This data is from only 3.8 million miles, in carefully controlled conditions with brand new high quality vehicles.
Besides this comes from Waymo, they didn't release the data used in their study.
Of course Waymo will try to paint their cars in the most favourable light possible. Maybe they are being honest, but that's not the point. The point is that public policy shouldn't trust the allegations of a private company at face value.
> Driving doesn’t require an advanced degree in physics. It’s mostly a mish mash of fairly mundane tasks.
Generally not wrong, but when it becomes non-mundane, it can become so in a hurry.
This is why I'm sometimes skeptical about statistics that say "x% of airplane crashes are human error". Well maybe, but by the time the humans get involved the automated systems have often thrown up their proverbial hands and digitally exclaimed "Jesus take the wheel".
I think having computers take over the mundane parts may help collision rates as people can get numbed and bored by the monotony.
One has to be careful about having the right level of automation and control for the given situation, as per the famous "Children of the Magenta Line" presentation:
> “Children of the magenta line” has become shorthand for pilots who manage and monitor systems but lack the stick-and-rudder skills to fly the airplane with authority. You may hear it wielded as an epithet in a discussion of “real pilots” versus “kids these days.” But in many cockpits today, automation illiteracy may be as dangerous as automation dependency. Vanderburgh’s insights shouldn’t drive us to reject automation altogether, but rather put it in its proper place in our pilot toolkits. Just as you should hand-fly regularly to keep your stick-and-rudder skills sharp, you should also practice using the technology available in the aircraft you fly.
I'm willing to believe that computers are better at lane keeping than humans are. That's not the only scenario, though: Unfortunately, driving can go from lane keeping to making a life-or-death decision in a second, and that's not enough time for a human hand-off.
A plane's autopilot can beep and bail out to a human with a few seconds for them to get oriented and take over; a car doesn't always have that luxury.
"Is this a person walking into my lane or just a shadow?" isn't a question that comes up in the air. Until a self-driving car can minimize the probability of that detection going wrong to below human thresholds, I wouldn't feel comfortable sitting in the driver (or maybe even passenger) seat of one.
It's rarely that straightforward: most all runway collisions involve ATC giving correct instructions, but the crew mishears them, misunderstands them, hears them correctly but accidentally goes to the wrong runway/taxiway, etc etc. And there's usually complicating factors like terrible weather, diversions to unfamiliar airports etc in the mix as well. What remains the world's worst single aviation disaster in Tenerife checked pretty much all these boxes.
> The ATC computer should talk to the plane computer.
Not as simple:
1) You need to implement it for very small airplanes too, which is far from easy. In the Linate air disaster a small Cessna Citation invaded the runtime due to fog and misscomunication.
2) You need to implement it for ground vehicles too, as in the LATAM example it was a fire truck that invaded the runtime.
3) You need A LOT of infrastructural update.
4) Even if you had computers talking to computers, this may still be far from enough to avoid collisions, as those mistakes are often a matter of seconds (as in the both accidents quoted before).
5) Automation brings its own issues. One of them is complacency where attention levels unavoidably get lower as reliance on automation increases. In turn complacency can make runways more prone to such incidents in case of malfunctions/bugs.
Yes, in this case, you do need to update everything at once. At least for a given aiport. All it takes is one plane, helicopter, fire truck, or baggage transport vehicle to screw with the assumptions of the system and jeopardize safety.
No, computers aren't necessarily faster, assuming we want to achieve autonomy on the ground. Note that the average human reaction time to visual stimulus is about 250ms. Achieving this on a plane would require a great deal of cameras and sensors that would not only need to stay clean, but capable of withstanding pressure changes, high velocity winds, the occasional unfortunate airborne debris, etc. Then there is the processing power to both process each input in real-time (per plane) and train it on real-world data to recognize a variety of scenarios to act upon (per model and perhaps configuration; AA alone has about 11 active aircraft types across 953 planes). Contrast this with Tesla's investment for approximately 6 models. [1]
Achievable? Sure. Cost-effective and sustainable with current tech? Doubtful.
It's hard to know what that is. TCAS is a great example of an obvious solution to mid-air collisions - two devices seamlessly negotiate for one plane to go up, the other to go down.
One I can think of is an electronic runway "lock". ATC mark the runway as locked to a particular aircraft. The landing aircraft checks at minimums whether if the runway is locked to any other aircraft. It's harder to know how to prevent incursions at ground level, but it could be a beacon near the runway threshold and intersecting taxiways, flashing a warning in the cockpit of the taxing aircraft if an aircraft on final holds the lock. A pilot could still disregard it. Sometimes two planes can share the same runway, such as a plane lining up while another vacates, which adds further problems.
Runway Status Lights (RWSL) are close to what you're describing. If surface surveillance radar detects that a runway is occupied, it turns on red warning lights at the runway approach end and at the hold-short bars.
The nice thing is that the entire system is ground-based, no additional equipment needed in the aircraft.
However, it's complicated/expensive, so only a handful of airports have it installed.
I think you might be way underestimating number 5.
While on the first hand your point sounds very reasonable, in practice it may have more adverse effects than we may think and make the whole system more unsafe.
No, the question is what authorities and regulators conclude, after this incident is properly investigated, regarding improvements and risk mitigation measures have to be put in place. Aerospace regulations are written in blood, and almost-blood, not in public outcry.
>1) You need to implement it for very small airplanes too, which is far from easy.
No, you don't. It's easy: just don't allow small airplanes at airports with large commercial aircraft. No one needs them anyway; they're just a luxury for rich people. The only place they make sense is for very remote locations, like Alaska, where there's a need for people to pay for carriage on such a small aircraft. For normal airports that carry jumbo jets, there's no reason to have small planes sharing the runways. For exceptions like coast guard jets such as this one, these are government-owned so they can afford the upgrades.
The absolute worst thing is for an aircraft to refuse control input from the pilots, regardless their state of informedness. Lest we forget, we saw what happens when computers take priority over pilots thanks to 737 MAX MCAS.
>Then let’s make it obvious and include it in the training.
Which cuts right to the heart of the B737 MAX issue - Boeing substantially changed the design, then added sensors & software to make the changed airframe handle similarly to the old 737 airframe and claimed that this meant the pilots did not need retraining
If Boeing had been upfront about MCAS, and required that pilots were trained in what it did and how to override it then the crashes probably would not have happened.
No, we need more than this: the system must be documented and pilots must be trained for it. If the aircraft manufacturer fails to do this, they go to prison. If it's found that hiding this information was intentional, the corporate executives and everyone else in the company who knew are publicly executed by firing squad (with them aiming at their stomachs).
Airplane taxiing is manual. ATC can tell the plane to use runway A, and the plane can check that A is free, but neither will stop the pilot from accidentally steering into runway B.
And yet, aviation is one of the safest industries mankind ever built. It seems those solution in place are actually pretty good.
Edit: Something I try to take to heart, if I don't have domain knowledge, I assume that whatever easy solution I can think of either was already discarded for valid reasons I don't know nor understand or is already implemented without me knowing about it. Generally, people working in a particular field tend to be competent in that field, just assuke that more often than not they know what they are doing.
We got very good at doing aviation safety in a certain way.
The people who work in aviation are extremely skilled at doing safety in a certain way.
This is usually true for any domain. People are very experienced and extremely skilled at doing things within a certain framework.
It can still be true that there is a better way. I realise it’s hard to be on par with a century of aviation safety experience, even if the new approach is better. That’s why you need a gradual transition that focuses on enhancement rather than replacement
I can’t speak to international ops because I’m just not sure. In the US though, we have runway status lights. At all the runway intersections there are lights embedded in the runway that will tell you if the runway is safe or not. A radar based system is constantly monitoring the runway and will light it up red if there is a potential for collision. The system works independent of any input from ATC. I think it’s actually quite effective.
Unfortunately it’s expensive and so only runways at the busiest airports have it so far.
And yet in US it's also standard practice to have airplanes takeoff on the same runway at the same time if the preceding airplane is small enough to not create vortices.
Note how no one did anything wrong really in this case:
>runway incursion monitoring and alert system had failed to detect the danger as it had detected that both aircraft were in the air despite still being on the ground.
Yea, as I mentioned it’s rather unfortunate that not all airports have it yet. I’d love to see it installed everywhere. Including non-controlled airports.
I doubt your “other domains” are as safe and reliable as the aviation industry. An automated system may be possible to make, but will require a huge undertaking to bring even close to the current safety standards, and would introduce its own safety issues
An elevator is many orders of magnitude less complex than an airport, though.
Total automation can be done quite safely in some domains. Between that and a completely manual/human-operated system lies the area of augmentation, which can both be a boon to safety if done correctly, or deadly if it leads to automation complacency or false expectations around the types of situations it can successfully handle.
In the past six years there was some accidents just in the EU, at least one decapitation happened in the elevator of an hospital what I know. Or in the city were I live, a fail in a badly lubricated brake when an old person was entering (open door) sent the elevator to the top, just a few weeks after passing the monthly -supposed- maintenance service of the manufacturer.
The maintenance seriousness in aviation is superior to an elevator, but even like that, the last human layer decision should not be avoided.
For example, bad behaving sensor feeding a fin balancing algorithm introduced by the manufacturer to avoid the need to redesign an unbalanced plane to accommodate larger motors it was designed for, and pilots unable to make the plane follow their commanding due such algorithm blind decisions priority... has crashed planes.
It's largely a solved problem in aviation as well. Serious accidents get a lot of attention specifically because they are so rare and those based on clearance miscommunication or other radio misunderstanding are rarer still. So you'd be introducing a lot of additional system complexity trying to chase after ever decreasing gains. I'm not sure the end result is a net positive for safety.
Human to human open discussion is superior to computer to computer in the business of flying/navigation.
It might make sense at some point in the future where an autopilot system knows how to taxi, take off, troubleshoot issues by itself during the critical phases of flight and so on.
And like others said, aviation is one of the safest industries on the planet.
A runway may be clear when given the clearance to land. And that happens about 4NM from the RWY threshold. In the meantime a incursion might happen. Given low visibility you may not see it as you land a big airplane whose nose is a bit high up because of the flare. And even if you see it, it takes a few seconds for engines to spool up to full power in order to do a go-around.
Not really. If an instruction is unclear the pilot will ask again for a repeat. Or if the instruments show everything to look normal but the plane doesn’t feel right, the pilot will correct it, a computer will not.
Think of the 737 Max whose computers crashed two planes because of incorrect sensor data being fed to them. The pilots there tried to rescue the planes while the planes were following their logic, which was flawed.
> If an instruction is unclear the pilot will ask again for a repeat.
Except when they don't, and confirmation bias themselves into doing the wrong thing. The most likely cause of a plane crash is pilot error, and runway incursions happen shockingly often [1][2][3][4]. The only reason there hasn't been a Tenerife-style disaster is sheer luck.
Look, incidents happen all the time in any industry. And for the five examples of human error you have thousands of examples where humans performed very well.
Computers are not infailable either. So while they add a lot of safety features and ease the workload of the pilots, they should not have final authority over pilot actions. 737 Max is a prime example of why.
737MAX is what happens when you have a bad plane design because of bad regulations, and corporate execs intentionally make the existence of a plane system a secret from pilots so that they don't need retraining. An incident like that should result in prison time, if not outright execution. It is not a good example of the downfalls of an added safety feature; the MCAS system wasn't a safety feature at all, it was much more like the emissions-cheating "feature" in VW's "dieselgate" scandal.
> Shouldn’t there be a universal standard to communicate clearances to the computer systems of the airplanes and other runway vehicles?
“Communicating clearances” would actually be the easy part. The hard part is making sure that there actually is clearance. The system would need to be able to “see” the runway and to “know” about the flights that are permitted to use it in real-time. It would involve many computers, some in the air and some on the ground I assume. If the system failed the pilots would have to fall back onto something else, perhaps what they do today, so they’d still need training and practice. If a pilot made a mistake the system would need to be able to detect that and update everyone.
I worked in software in a domain vaguely related to ATC, and I've heard a few years ago that even sampling the voice commands (and routing them as VoIP) is problematic , as altering the signal through sampling and encoding can go against regulations. I don't remember the details though.
It was in a project related to ATC and area control / FIR services backup systems.
Pilots are accustomed to the immediacy of voice communication, which any electronic system will struggle to replicate. More sophisticated radio than simplex VHF has too many variables, and agreeing on an international standard would be challenging.
There have been some developments of a replacement of, or at least a supplement to, simplex radio. But none are as cheap, quick and reliable as verbal communications with readback.
Assuming (1) the replacement is a better solution, and (2) it is feasible to retrofit every aircraft in the world and switch everyone over to it.
Just in relation to (2), I can't really think of anything worse than having two communication systems on operation at the one airport - everyone must be using the same system.
Nor will it be a good argument in case of a fatal crash to say "well, it did contribute to this terrible accident, but in the long run it'll be better for everybody!".
Nobody in aviation is conservative about automation due to a lack of a desire for safety; quite the opposite. But it's just a fact that e.g. an automation system that can prevent a type of human error entirely in 99% of all cases can lead to more accidents in total, if it means that humans aren't regularly exposed to these 99% of cases which in the end train them for the 1% that the computers can't handle.
And this is, as are so many things in aviation safety, learned through blood.
You seem to have a pretty strong but not necessarily very substantiated opinion about the topic. I really suggest you read a bit about it, if you're so inclined. E.g. Admiral Cloudberg has been mentioned in this thread, it's a very good resource of accident recaps and the impulses they give for regulations and developments.
In contrary to what you're implying, "we've always done it like this" is not a principle of aviation safety.
Indeed you'd be hard-pressed to find any other globally coordinated system that is so dedicated to and successful at the gargantuan task of continually improving itself, a trillion dollar industry, and regulatory bodies all over the world. A five decades long track record of drastically reducing fatalities in the face of rising passenger miles speaks for itself.
To put it a bit bluntly, I'm pretty sure they don't need to be told that it would be "just as simple as (...), because that works for an elevator".
I think the problem might not be super simple to solve through updating vehicles.
You need to update infrastructure globally (and thus you need coordination between multiple different agencies), then you need to implement the solution at both airport and vehicles level.
That's a lot of work. In case of LATAM the Airbus hit a fire truck (which would've too required such a solution). In case of the Linate air disaster, such technology would've had to be implemented on a very small Cessna Citation model.
Maybe the updates need to be implemented at the airport level alone, quickly signaling controllers of possible violations/conflicts, and in fact most major airports have both ground radars and clear ground signals to avoid these incidents.
But it's still impossible to avoid these issues as these errors may happen in a matter of very few seconds thus not giving enough time to react to any of the involved parties.
How does the incident/accident rate in these other domains look like, then?
> How on earth does all this still depend on skilled individuals not making a mistake?
Everything safety critical always depends on skilled individuals not making a mistake, as long as there isn't full automation (in which case it depends on the people designing and implementing the automation system).
Until we do have that, mixed automation can actually make things less safe if it isn't done very deliberately and carefully, with people in operations relying on an imperfect system, and system designers on ops to catch complex and edge cases.
I agree. It’s SO antiquated. Nevermind issues with clearance there’s even issues like pilots forgetting what was ordered or ATC not catching incorrect read backs. This is just ridiculous that still in 2024 we are using technology from 1950 as the only way to give instructions. We solved encryption and authentication years ago. FAA should mandate strongly authenticated computer messages as the primary.
But you know the real reasons it’s old.
1) corporate greed and no regulation
2) legacy… it’s always legacy holding us back. All the third world countries, Russia, etc would not adopt it.
I wouldn't blame Russia for stupidly refusing to upgrade to newer technologies. Russia has its faults to be sure, but I'd sooner blame all the "freedom-loving" general aviation enthusiasts in America: you know, the people who refuse to give up carburetors and leaded gasoline in their 1940s-era airplanes?
Shouldn’t there be a universal standard to communicate clearances to the computer systems of the airplanes and other runway vehicles? And shouldn’t there be a small, well-tested core system that makes sure clearance never conflicts?
How on earth does all this still depend on skilled individuals not making a mistake?
We got over this in other domains long ago.