Aviation is stuck in the 60s – a reflection on MH 370
english.martinvarsavsky.net
english.martinvarsavsky.net
Planes have GPS trackers. Not only their company knows where they are, the control center can too. In the case of the north atlantic track system, air control keeps a tight eye on speed, altitude and separation with very precise measurements, even as airplanes are far away from land.
I just fail to see the point of this post. I recently hitched a ride on the cockpit jump seat of a modern airplane for a Europe-East Coast flight and during that I saw the air traffic control knowing exactly where we were, and the pilots communicating via text message (ACARS) with control, as well as using satellite links to contact dispatch (via text messages), as well as satellite phone calls.
Pilots keep paper around them because pilots are there to maintain control, and paper is just another failsafe (with pretty good reliability record!).
Bonus: some planes can notice alterations in the flight dynamics and report an ice buildup. No pilot in this planet is going to let any external person input flight parameters remotely into their aircraft's system while they are in the air.
https://en.wikipedia.org/wiki/High_Frequency_Data_Link
It's actually a brilliant system, with network of transmit and receive sites its like globally available text messaging for pilots.
Also, unless the pilot was using atrocious equipment I can't see why HF radio would sound awful. I use my HF radio almost daily and don't have any complaints. I can legally listen to air traffic controllers on HF with no problem as well. The equipment I have is far, far inferior to that in most commercial airliners.
[1]http://www.maxtrescott.com/max_trescott_on_general_a/2012/06...
Haven't you read about the missing aircraft?
Malaysia Airlines Flight 370 is a missing international passenger flight operated by a Boeing 777-200ER aircraft: http://en.wikipedia.org/wiki/Malaysia_Airlines_Flight_370
At least two of the passengers were travelling using false identities (stolen passports, tickets bought at the same time).
Edit: last known position: http://www.flightradar24.com/2014-03-07/16:46/12x/MAS370/2d8...
Or other examples that manage to disable enough of the flight controls but don't immediately break up the plane
It could be hijacked, landed, and held hostage, but it's incredibly difficult to hide a 777 anywhere in the world where it can safely land. It's possible it was hijacked with intention of being held hostage but crashed, hence no claims of responsibility.
This way, most problems an engine would develop that would be hard to spot by ground service, are detected and acted upon long before they threaten the operation of the engine and airplane.
There are planes and planes. Security in airliners and in light aircraft are two matters of their own. You don't expect people who own a Cessna to have a black box and a satellite data link.
sure you can create an automated dispatch system on an iPad. How do you authenticate it? how do you tell if its not working? does it fail safe? Making safe software is hard, and well beyond the wit of your standard programmer.
After all, can you gaurentee your software when the CPU is at 100%? can you say with certainty what happens when your CPU is hammered by all the interrupts at the same time?
Everything in the whole stack has to be verified. Thats means no virutalisation, no ruby, no perl, no python no ethernet. You can have firewire 400 though.
One can almost guarantee that this plane crash did not crash because of a failure in ATC<->Pilot communication.
Those shipping container trackers use GSM to call home. They're cell phones. What happens when that container is in the middle of the Pacific Ocean?
How does a plane get data back to the ground when it's 7 miles up and 1000 miles from shore? You have to have a satellite or long-range VHF link.
For those interested: http://en.wikipedia.org/wiki/Automatic_dependent_surveillanc...
[0]: http://dangerousprototypes.com/2013/02/28/defcon-20-renderma...
those 3 should probably be added to current planes, regardless.
i'm not sure why you think there is a single system with a "CPU" in a plane either. there's multiple independent systems - some electronic, some analog.
all this not much to do with visualization or python, in fact... and software glitches in planes do happen, by the way.
That means that it has to go through actual real QA (something foreign to modern Software science.)
You want to put in a glonass receiver? sure, but it has to operate independently of the GPS system, and it cannot replace any other navigation system unless the FAA allow it.
the point is each system must be robust (i mean really robust) and able to work past the extremes independently.
In the era of having artificial hearts and pacemakers that are helping better lives, I don't think its as hard to even build sound and safe automated flying systems.
Really? When you have to worry about things like "Squelch" in order to deal with voltage supply ground noise? You are on thin ground.
The real problem with modern technology is that the pilot can't rely on it. Sure, a digital radio is $1,000 (or less). However, the digital radio that is FAA certified will be expensive and nobody is going to develop it because the FAA won't require it because all the pilots will bitch that it is too expensive.
But nobody is going to front the money to get it FAA certified so that it can be a required solution given that the most you will sell is 1,000 a year.
This is the real reason why air technology is so far behind. There is no profit in it.
When speaking about systems, which need maximal reliability and are strictly controlled by authorities (air planes, nuclear power plants, medical treatment devices, space related stuff), updating existing and once approved systems is so painful, that you often do all you can to avoid it. That's a sad state of affairs.
author's point is that we could do better using today's technology.
I'm not saying that things couldn't be better (and I'm no expert at all) but I doubt that any of these new technologies would have saved MH370.
It's very weird, for instance, that the ground station that reports to flightracker24 lost ADS-B communication with the plane all of a sudden and it's pretty clear from the number of planes that you can see in the area that it is probably not a coverage issue.
It seems that Flight 370 had GPS tracking (flightradar24.com screenshot): http://www.voanews.com/content/vietnam-navy-official-missing...
Direct link: http://www.flightradar24.com/2014-03-07/16:46/12x/MAS370/2d8...
All bets are also off when the plane has had a catastrophic loss of power.
First, I agree with your thesis. In fact, I'd argue General Aviation is stuck in the 1950s. Lots of reasons for that, including the cost of insurance.
Second, I had a hard time reading this, as you seem to lack the ability to form cohesive paragraphs. Very sorry to have to tell you, but I figured somebody should. Nobody can read a big wall of text where each paragraph is trying to say several different things at the same time.
Best of luck in future drafts.
I ask you to envision the logistics of a simple change. You want to implement X, and gee-whiz do I have some cool, new technology that makes it easy!!
Okay. Let's start. If we make mistakes people die, and careers end. I'll let you draw your own conclusions about motivations re that, but all I'll say is as an engineer I only ever focused on the former. So, huge testing and verification effort to show that my gee-whiz technology that works in some consumer device has acceptably low failure modes, does not negatively affect pilot work flow, can work under the conditions of flight (-55C to 70C anyone?), over the wide performance characteristics (high G, high vibration, salt water, high radiation), and so on.
Okay, you did all that! Let's start bolting it on the planes!
No.
Let's write and get approved training programs for all the maintenance workers. Then, let's train them - across the whole world. Let's write and get approved training programs for the pilot. We will fit it into their refresher training, have new mandated training, or what? Basically, rewrite all the training curriculum that is out there. Get it into the schools, so the pilots coming fresh out of school aren't behind. Again, across the world.
Oh, this interfaces with the towers? Okay, so do all that again with the towers. Hmm, you want this 2013 technology to seamlessly integrate with some core memory technology - that should be easy. Perform a study, put out RFQs, get bids, select the best bidder, have them build the system, manage them through the cost overrun, opps, 3 months before deployment Congress mandates that that core-memory system be retired, and oh, how will this work in the 168 other countries?
Got that sorted. No, wait, no one in the tower knows how to use it, no one knows how to install it, no one knows how to maintain it. Let's throw money and time at that! Oh, unions. I hate unions. ATCs have a union. This could take awhile....
Finally, it is 2020, and I am rolling out, um, 7 year old technology that is entirely obsolete and no longer supported by the manufacturers. Oh, they'll support it if you throw enough money at it - get your $5 microprocessor at $1000 a pop.
Meanwhile, the entire world is filled with aircraft still using the old system. So, we mandate a phase-out by 2035. Just another 15 years of supporting the old and new systems in parallel. I'm sure that'll be pretty cheap.
People who work in the field will rightly accuse me of hand waving, and especially of over-exaggerating some difficulties (not every modernization project hits every possible snag that exists). But this is still a useful sketch the scope of the problem. I've spent time talking to very high people in the FAA. They are not unaware of the old systems and their limitations, nor are they bumbling bureaucrats (pet peeve - it is easy to villanize faceless people, and that is very lazy thinking). We in industry are forever proposing new ideas, better technology, and so on (let's face it, they are all trying to feed at the trough of government spending, and getting your system mandated is a company maker). But the price tag for my handy,dandy system is at the noise level compared to the cost of the logistics of deployment.
I am not arguing that there is nothing to be done, or that everyone is working maximally efficiently right now. Certainly the US is behind other countries in some areas of aviation technology. But it is not in any way a trivial problem, one of "just bolt a new radio to the plane and trash the old ecosystem".
edit: consider, for example, the Rockwell Collins DTU-7000 Data Transfer Module (https://www.rockwellcollins.com/sitecore/content/Data/Produc...). This is absolutely modern hardware in the aviation world. It is PCMCIA. And how exciting it was to get. You would not believe the cost and size of the old system - we would jealously keep tracking logs of, I forget, a few MB of flash memory units that cost thousands and thousands (and thousands) of dollars. There is some even more modern stuff being rolled out that uses usb. But consider, when this is something that contains your flight plan, your maps, and so on, the cost of a stray gamma ray blowing away a byte. Mull on how much testing this hardware goes through. And then factor in all of the logistics above. We already don't have money to own the old system, and now I have to go to all this further expense, to save what is truly chump change (that thousands and thousands and thousands number) in the end? Millions to save thousands.
Of course, we have to modernize, we can't store rich maps on tiny memory, so we spend, and spend, and spend. And then get a front page HN story about how old everything is! Well, there's a reason for that.
It's almost like this blog post is suggesting that safety of large airliners is in some sort of massive disarray that could be fixed by an iPad and a few apps.
The second thing that we fail to think about is statistics. Systems like hadoop are very popular because is because failure ALWAYS scales, so we just buy lots of things and assume the embrace the risk of failure. For commercial plane, even a 0.1% failure rate would affect 25 flights out of O'Hare every day. Square had some information that 10% of customers that used an iPad 2 had one fail within a year.
So, the idea of expediting unproven technology with unknown failure rates to a system as reliable as a 777 sounds utterly preposterous to me.
Even if there was no bureaucracy everyone involved still has to go through rigorous training before we get to the point where "it's not difficult" ...
Aviation is a field where every major incident, and many minor ones are investigated and root cause is determined. The FAA and NTSB does fault tree analysis of crashes and publishes lots of material and bulletins to prevent similar issues.
What happens when the GPS says one thing, and the "internet" says another?
what takes precedence?
The "Internet" isn't safe, reliable or even bullet proof.
New technology isn't a golden bullet. Often its just a re-invention of the wheel. (whatsapp for example, is just MMS without the guaranteed delivery.)
I don't want my plane being crammed full of shiny new tech, especially if its not proven. Th reason why there is bureaucracy is to make things safe.
As for preference, it's already norm that there's a multiple different input sensors, and the autopilot and the real pilot has to deduce the correct value. (Good example is air speed, see Air France Flight 447).
Now imagine that everybody on the alternative secret phone system is flying around hundreds of passengers who'll be killed if somebody misdirects a call with important instructions.
I've found that, in general, if one's reaction to a practice is "WTF, is literally everybody else in this industry stoopid?" it usually means the opposite...
MH370/MAS370
Boing 777-2H6ER
Registration 9M-MRO
Altitude 0ft
Speed 471 kt
Track 40°
Vertical Speed 0 fpm
Lattitude: 6.97
Longitude: 103.63
Radar: F-WMKC1
Squawk: 2157
That particular airplane was delivered new to Malaysia Airlines in May 2002 and was involved in a ground mishap in 2012. While taxiing at Shanghai's Pudong airport, its wingtip hit the tail of another aircraft. According to an independent accident-tracking site, the damage suffered by the Boeing 777 was "substantial.": http://aviation-safety.net/wikibase/wiki.php?id=147571
"Malaysia Airlines confirmed that the missing aircraft had been involved in a collision with another plane in 2012 at the Shanghai airport that resulted in damage to the Malaysian aircraft’s wingtip. But the airline said the wing was repaired by Boeing and declared safe to fly."
Pilots get weather forecasts before take-off, are they inadequate? How does knowing the weather at the destination help with the immediate task of flying a plane through a thunderstorm that you knew was going to be there anyway and you can see on your radar?
How more weather information change the impact of a major weather event e.g. East Coast winter storm, on air travel? I don't see how it could, the decisions are made far ahead of time because weather forecasts are pretty good.
Purposefully staying behind the technology curve is not a "safe" strategy, aviation is not alone in this obsolete thinking bias, the same is seen in utility, automobile industry too.
http://www.telegraph.co.uk/travel/travelnews/8111075/Live-mo...
Today's airliners are incredibly advanced. In the 60's they didn't have Terrain Collision Avoidance. Try to fly the plane into a hill and alarms go off and tell you how rapidly to ascent. They didn't have automated traffic collision avoidance (TCAS). Fly two planes at each other and it tells one pilot to dive, the other to climb and how much. They didn't have stick shakers which alert the pilot if the plane is near stalling, before it stalls, so they know to gain air speed to prevent the stall.
Those are the basic standards. Today any aircraft worth it's salt has a full Doppler weather radar to avoid storms, hail and down drafts. Today the entire flight path is computer calculated and tracked from end to end. Just turn on auto pilot or keep the indicators aligned on your artificial horizon and you'll get where you're going. No one gets lost these days without trying. I've not going to even start on CatII and CatIII landings too. Today planes can land in 0, that right 0, visibility.
> if it wasn't airliner stuff would all be automatic
But let's keep going. Ever heard of AutoLand? Yep, plane can land itself. Ever heard of AutoFlare? Yep, plane will tilt back at just the right angle for a pillow soft touch down. Ever heard of AutoThrottle? No need to keep adjusting those engines, plane takes care of that, just tell it where you want to go. Frankly, pilots rarely do much more than tell the plane what it should do. Want to go to flight level 380? Just turn the knob to 38000ft, set the desired vertical speed rate and the plane will adjust the throttles, ailerons and trim itself out at 38000ft for you. Want to go to heading 245? Just dial in 245 degrees the plane will take your speed into consideration and speed up just enough to keep everything constant through the curve.
Funny thing about automation though is not getting everything automated, but keeping pilots from getting stale because too much is automated. Take the Asiana flight, it's what we call Controlled Flight into Terrain. Not a thing wrong with that 777 but the pilot.
We're worlds away from the 1960's and the 777 is among the most sophisticated and advanced craft in the world. Far more advanced than even our spacecraft of the 1960's.
But TCAS and GPS-aware Enhanced Ground Proximity Warning System (EGPWS) are 90s, and automation vs maintaining flying skill is definitely a major challenge.