Hacker News Reader

Comment by cqqxo4zV46cp | Hacker News Reader

I’m just an avid ATC listener. I’m not a pilot, air traffic controller, or anything of the sort. I could very well completely off base. With that massive disclaimer out of the way, some of these changes really do come across as “we need to change something” meddling.

View on HN

There's the whole Ishikawa situation going on taking up the news in Japan, so I think there's very little pressure for them to just appear to be doing something.

On the other hand, they might be suspecting something, and apply this change as a stopgap just in case, even if they later settle down on another cause.

When dealing with incidents where there is no clear root cause, it's often advisable to implement guard rails to reduce the likelihood and/or impact of a repeat occurrence. As a result, this seems like a sane change.

I couldn't tell in this case, but sometimes it is also definitely useful [0]

[0] https://en.wikipedia.org/wiki/Tenerife_airport_disaster#Lega...

Pretty much. If you have human controllers and human pilots, eventually someone's going to misunderstand an instruction. It happens every day all over the world. The backstop to this really can't be "give different instructions", because whatever you pick can be misheard too.

Obviously the primary backstop is supposed to be the pilots. The Dash 8 should have been listening to other traffic on that control frequency and should have known that there was an aircraft on final for that runway. The pilots of the A350 (and I don't understand why this angle isn't getting more attention) should have been looking out the window and seen the lighted-and-very-visible turboprop on the runway in front of them.

> The pilots of the A350 (and I don't understand why this angle isn't getting more attention) should have been looking out the window and seen the lighted-and-very-visible turboprop on the runway in front of them.

Because pilots know that it is incredibly hard or impossible to see an airplane lined up on the runway.

By lighted we mean that the turboprop had the required lights on the end of the wings and the end of the tail. Otherwise the airplane is totally black, on a totally black runway. Simply it is very easy to not notice the extra lights among all the lights of the runway.

The pattern you are trying to notice is 3 extra lights among many, and a small gap in the runway lights where they are obstructed.

> The backstop to this really can't be "give different instructions", because whatever you pick can be misheard too.

We know a solution for this. ASDE-X(Airport Surface Detection System) is a set of sensors (radars mostly) integrated together to monitor the ground movements of aircraft and automatically report deviations such as runway incursions.

https://www.faa.gov/air_traffic/technology/asde-x

Have you ever flown a plane? You're essentially blind to objects directly in front of and below you.

That's also true in a car of course, but there's a lot more ground below you when you're flying a plane.

From my experience, people who don’t know what they are talking about love to talk about pilot error and assign blame to the humans in the cockpit. You don’t see that often when reading experienced commenters who know that there should be 10 safety systems and features to prevent mistakes from getting catastrophic. Admiral Cloudberg’s analyses for example discuss at length the failures of all the mechanisms that allowed the crash to happen. It’s never “pilot need to git gud”. (Well, almost never, it was very nearly that for the recent Yeti Airlines crash).

It doesn’t help that there appears to be a, “cultural thread”? in aviation which believes that it’s a sign of your chops to talk like an auctioneer on the radio. There’s an infamous example on YouTube somewhere of this ATC telling pilots “cle te leh” for Cleared To Land.

If the signalling system is sufficiently modern, a train cannot drive over a red signal without being shut down. You cannot turn a signal green if there is another train on the track.

Why do we have this kind of state management for trains, but not for airports?

Because you cannot stop a plane mid air to wait for confusion to be cleared up.

No, but you can sound a loud warning in the cockpit if you're lined up as if on final and the runway isn't clear. Go-arounds are a standard maneuver.

It's dismissals like this that are preventing progress. I think the upthread comment was insightful and correct.

How do you determine programatically if the runway is clear? Or more accurately if it won't be clear when you get there? That's a substantially harder problem than you're letting on.

So, just to be glib, my car reports GPS positions at about 3 Hz at all times to an app (Teslamate) under my control, which is far more accurate than needed for runway incursion detection. And to boot, it literally drives me around, autonommously, every day, in an environment (my suburban neighborhood) vastly more complicated than any mere airport.

To say that the problem will require some engineering is correct. To claim it's infeasible is just ridiculous.

With a 'runway incursion system' - https://www.atgairports.com/solutions-and-systems/airfield-c...

You don't necessarily need a system that scales all the way down to tiny airports.

Major airports like Haneda presumably already have things like stop-bar lighting, for operating in foggy weather. If we already know where each aircraft is, and whether they're permitted to cross the stop bar right now - I can believe this runway incursion could have been detected.

Haneda does have such equipment, but it was not working at the time. This fact was promulgated in a NOTAM in effect at the time.

https://www.businessinsider.com/japan-jet-crash-runway-warni...

Well that at least seems like relatively low hanging fruit.

It’s hard, but come on. It’s totally doable.

> Or more accurately if it won't be clear when you get there?

I love how all the replies to your comment completely ignore this point.

If implemented as the other commenters naively suggest, it would cut the runway acceptance rate by probably 90%. Planes are normally sequenced so that each one touches down just seconds after the previous one turns off the runway.

Otherwise, the horn would be constantly blaring for the entire approach, which would do exactly one thing: train pilots how to ignore the horn. Exactly like gear warning horns in planes without a radar/lidar altimeter (based on airspeed and manifold pressure). So many gear-up accidents happened with the pilot reporting that he was distracted by the warning horn on final. Why? Because there were so many false positives that the horn became nothing more than a nuisance device. And now you’re in exactly the same situation as before the incursion warning system, but the pilots on approach have to communicate over the incursion horn all the time. Great idea, guys! Why didn’t anyone think of this before? Those aviation folks sure are stupid, aren’t they?

Even more problematic, the coast guard plane entered the runway just seconds before the JAL flight touched down. Even with a warning system for a plane on approach, there probably wasn’t enough time to execute a go-around.

Indeed. I am definitely not an aviation expert or professional but I do have a pilot's license and some hands on familiarity with how this all works.

It's not easily automated. That's just factual.

It's not impossible but it's way trickier than people who have never thought about this much are letting on. The accuracy of GPS is absolutely irrelevant for example.

On topics like this there are always "why don't they just do this" comments as if nobody in one of the most safety-minded has thought of it yet. I collect them in my HN favorites. Also there is a lot of safety that comes from not being too reactionary and adding complexity anytime someone has an idea. I think it's better to ask something like "does the industry use or plan to use any automation for this?"

You are right, designing a perfect system is complicated. Of course, you have account for planes that are expected to leave the runway before the next touchdown. The train equivalent for this is called CBTC. https://en.wikipedia.org/wiki/Communications-based_train_con...

My bigger point here is that there are a bunch of low-hanging fruit that could have prevented this. A system doesn't have to be perfect to be useful.

The tower could have known, that a plane was expected to land in the next 30 seconds. They could have automatically transmitted that to the cost guard plane and this could have sounded an alarm in the cockpit of the taxiing plane.

You 100% certain your system won't false? Playing a loud horn/sound on final is a good way to overload the crew and needlessly cause a crash.

If thee crew is this easily rattled, how come we trust their communication skills so much?

I see your point, but I think that the rate of false positives can be tuned to be very low, with a good design of the system.

In the rare event of a false positive, it’s still extremely unlikely that this will cause any more disruption than the co-pilot pushing a button and double checking with the tower.

I never said the crews were easily rattled. I can see you're more interested in argument than discussion so I'll leave you with this thought: software engineers can throw features into a product and A/B test it. When aviation engineers throw features into a cockpit without thinking through EVERY side, accidents happen. You're technically correct that, when skies are clear and the crew is idle, adding sounds and buttons is no big deal. But that's obvious, and not the scenario you need to be thinking about.

EDIT: I thought it was obvious from context but maybe not... Imagine I said "overload an already heavily loaded crew." An unloaded crew isn't normally worth worrying about.

I'm not sure a properly-designed warning would add much load to the flight crew.

They're already listening for "go around" commands from ATC. This would just be an automated "go around" based on runway sensor data.

The aeronautic industry settled in communication to prevent accidents, the tracked transport industry settled on physical lockouts. The lack of a solid, stable pathway probably influenced the path of the aeronautic industry's safety protocol development.

You can combine both.

Trains fail safe. If you cut the power to the wheels and apply the brakes, it will be inconvenient, but generally fine.

Planes by their nature fail unsafe: if you cut the power they fall out of the sky and crash.

So it's more difficult to design a signalling system for planes. Still, the idea seems good: couldn't we have a system that detects if the runway is occupied and broadcasts that information on ILS?

> Still, the idea seems good: couldn't we have a system that detects if the runway is occupied and broadcasts that information on ILS?

https://en.wikipedia.org/wiki/Airport_surveillance_and_broad...

Forget air traffic control - who new ground control was so difficult and fraught with danger.

I happened to watch this video last night, which goes into some of the details of RIMCAS, the Runway Incursion Monitoring and Conflict Alert System, and how it failed to alert to a dangerous situation a few years ago:

https://www.youtube.com/watch?v=nj7nG6gJqsU

Reply on news.ycombinator.com