Aircraft Cockpit Automation - Good or Bad

Marc

Fully vaccinated are you?
Leader
a Pitot Tube Air Speed system is say $30K or more, to heat the same is ?? additional $$.
The pitot tubes were certified in iceing - I don't remember the exact combination of events and atmospheric conditions in which they froze up. Making things worse was it was a night flight and they were over open ocean, and if I remember correctly there was a storm (not sure of the weather part). I think there was a redesign of the pitot tube, but all of them are heated even ones on small airplanes.

Read Air France Flight 447 - Wikipedia for the details.

As is typical of air disasters, a number of "little" things which together doomed the flight. It is very rare that a single event leads to a crash.

where instruments are present one can always use others to cross check...
Well, usually, but not always. And think here... Time is short, or may be short, you have a number of instruments, some redundant such as the attitude indicator which many large aircraft actually have 3 of - Which is right? Which is wrong? This is another "it's simple to say" thing. Like I have said - In a simulator, or on a training flight, everything is in control so it's simple to say "...if this happens, do this...".

I remember one check flight I had where the instructor did an engine failure, he throttled the plane to idle and instructed me "handle it". We were over farm fields, I lined up with the rows, gear down, full flaps, and was all set. I was young and cocky, and the check pilot was a guy I didn't particularly like. He had told me "do NOT touch the throttle until I tell you OK". So, I took him at his word. I'll bet I wasn't 50 feet AGL when he freaked out and said "what are you doing?" and opened the throttle to full power. I had been taught the check pilot's word was that of god. By reaching 100 feet AGL he should have said something like "OK - Full throttle - Good job", but hadn't. He said nothing until his freakout. I was ready to set the darn plane down in a soybean field in Missouri.

The flight was a "qualify in type" check ride. He didn't even have to pull that check, but he decided to.

Ah, the good old days. Back then it wasn't even a big deal to land a smaller plane in places like St. Louis or Kansas City. These days that's pretty much a no-no unless there's an emergency. Back then I was doing some charters and "mail" runs between Cincinnati, KC and Omaha. I usually hit St Louis for fuel and sometimes KC. I remember one night coming into KC - Middle of the night, a somewhat foggy night and I had spilled some coffee. I couldn't spot the runway - So many city lights. I was talking with KC approach control - We were both tired and cranky. I was close, maybe 10 miles out. I asked him to turn up the runway lights. He did a bit but I still couldn't make them out. I asked a couple times again. Finally he was frustrated and turned them on full. That runway stood out like a sore thumb. I will say all the various controllers I worked with, so to speak, were wonderful people. Helpful, professional and just great people.
 

Ronen E

Problem Solver
Moderator
When I learned to fly years ago the mantra was "always trust your instruments". In general it is sound advice.

It is sound advice because today it's far more likely that the instruments will do a better job than a human. Not that instruments don't fail, but at what rate do they fail catastrophically?... Again, the problem is that people measure technology performance against an ideal 0% failure rate - unrealistic. If one wants 100% confidence they won't be in a plane crash - don't go on a plane.

The focus should be much more on building technology that can cope with multiple simultaneous failures, regardless of where they are - engine, sensor, computer or else.

AI can already teach itself best performance by running countless simulations / trials in very little time: AlphaGo Zero - Wikipedia I'm quite sure such AI system can quickly become better than any human in responding to any conceivable failure combination.
 

Ronen E

Problem Solver
Moderator
As long as driverless cars share the roads with cars with drivers, there will be problems. A coworker was bragging about his new car that automatically applied the brakes if the car ahead slowed down. A week later, he was complaining about the same feature because another driver cut across three lanes of traffic (Chicago) and caused his car to brake suddenly and hard. A human driver would have recognized the type of behavior and either let up on the gas or braked gently until the car passed across the lanes. Even if 100% of cars are driverless, how well will they handle black ice, wind shear, a flooded road, etc.

I generally agree that it'd be much better when all cars around are autonomous, but it could still work very well with human drivers around. Just a matter of time and work. This technology is still relatively young and still evolving. I believe that with enough sensors and computing power, an autonomous car could do just as well as a human - no matter what conditions and situation. This is especially true if you introduce AI, however there's a big debate on the wisdom in doing so (with non-negligible amounts of fear involved - justified or not is another question).

The same thing applies to planes. Software is great for routine, repetitive situations, but there will always arise a situation beyond the softwares programmed limits. AI isn't here yet.

It's getting there, maybe faster than most think...! And the beauty of it is that the programmer doesn't need to think every possible scenario upfront. Either way, it will eventually (quite soon IMO) outperform any human operator of any kind, so even if not perfect it will be much better than the alternative.
 

somashekar

Leader
Admin
Good for you. As if human drivers (all of us included) don't have any bugs.
Doesn't it sound very ridiculous......that human have all the intelligence and the bugs and it is used to create something that is set to replace humans.
Reflect upon this. No one has understood nor exploited the capabilities of the human brain to an extent of replacing the human itself. People who manage technology, if they cannot balance the human and the automation interface are setting up curtain raisers of massive disasters
 

optomist1

A Sea of Statistics
Super Moderator
think for a moment, how often one experiences issues with your cell phone, laptop...etc... the rush to fully embrace and implement AI, increased cockpit automation...if fraught with many seen and unseen pitfalls. When technology, fails as it frequently does, there must always be a method (a system/device defeat button) to override it and let the human(s) take control?
 

optomist1

A Sea of Statistics
Super Moderator
Sorry folks not to throw an accelerant on the embers of a good healthy current and relevant discussion...but the following quote attributed to Burt Rutan, seems to have some relevancy here: "For decades, as a professional experimental test engineer, I have analyzed experimental data and watched others massage and present data. I became a cynic; My conclusion – “if someone is aggressively selling a technical product whose merits are dependent on complex experimental data, he is likely lying”. That is true whether the product is an airplane or a Carbon Credit."
 

Ronen E

Problem Solver
Moderator
there must always be a method (a system/device defeat button) to override it and let the human(s) take control?

...Sure humans will always take the right step to remedy in such a situation...?

Example (completely imaginary):
Nuclear reactor temperature automatically controlled.
Control system fails.
Temperature rises and approaches critical limit.
Human operator panics, disengages control system and applies maximum cooling.
Excessive cooling triggers unforeseen reaction.
Reactor fails.

More appropriate response: Redundant automated control system recognizes that at high probability the primary control system has failed, disengages primary system and takes over, applies moderate and appropriate cooling, day saved.

Automated systems aren't afraid of dying (well, yet) so less likely to panic when at risk to self.
 

Ronen E

Problem Solver
Moderator
Sorry folks not to throw an accelerant on the embers of a good healthy current and relevant discussion...but the following quote attributed to Burt Rutan, seems to have some relevancy here: "For decades, as a professional experimental test engineer, I have analyzed experimental data and watched others massage and present data. I became a cynic; My conclusion – “if someone is aggressively selling a technical product whose merits are dependent on complex experimental data, he is likely lying”. That is true whether the product is an airplane or a Carbon Credit."
Who's selling anything?... We're a bunch of individuals exchanging general knowledge and opinions.
 
Top Bottom