Would be interested in a video about automation dependency and the kinds of incidents/accidents we've seen as a result of it! Nice video by the way keep up the good work!
@@AirspaceVideos it's s delicate subject to cover because it's easy to report on the bias rather than the actual effects. What I mean is, that if automation saves you it's not as news worthy as when automation fails. If you cover it I think you'd need to compare cases that are completely connected. So here the alpha prot caused a crash but you'd need to give examples where alpha prot prevented a crash and you'd need to give numbers and statistics. It definitely is an interesting topic but it's super delicate as well and needs a huge amount of research and i doubt that it would fit into a 10-15minute time frame tbh.
Full automation is the only way to go. This incident then wouldn't have happened - because the automation would be so thoroughly checked, double checked, triple checked and quadruple checked by independent inspectors within the company and likely one from another company - like an insurance company.
This is something that has always worried me. Having lived in Wellington New Zealand I, as a passenger have been very aware of the extreme and often fairly unconventional techniques required to get planes down on the ground in these sorts of conditions. These techniques usually involve keeping high levels of trust on right up to touchdown. I am really not sure how they do all this stuff in a modern plane. I would like to hear your thoughts on automation and iit's a substitute for actually flying the plane manually.
I think automation is great, overall. 99.9% of the time, it works just as intended, and that's great. I wouldn't want to fly a transatlantic flight fully by hand, that would be insanely tiresome (and boring). There are also times, where automation can help greatly. For example, when I fly an approach in very gusty weather, I like to delegate the task of maintaining airspeed to the autothrust. It does a great job of keeping the appropriate speed and I have more mental capacity on getting the plane's trajectory just right. However, this does not free me from controlling if the autothrust does it correctly, and I must always know what it does. If I don't, I switch it off. However, no system is infallible, and like that, accidents like this one happen. It's very unfortunate, but also very very rare.
@@AirspaceVideos I like your answer, the thing I pick up from it is that you are using the automation to help you fly. However, you are not letting the automation control you.
exactly! These are two of the golden rules of automation usage - know what it does at all times and take control if it does things that you don't like.
I like to see it as cruise control in a car. If you turn it on, it takes over and maintains speed so you can focus on other things, but at no point does it remove the ability to control the car from you. If you tap the brakes, it disconnects, and if you accelerate manually, the cruise input and your input are effectively added to each other. If you hit autopilot disconnect, same thing happens, the autopilot disconnects immediately and you are now flying the plane.
I think automation can be a great thing, but one of automation's biggest dangers is the designers thinking they know better than the pilot. I don't know this system well, but I'd say that it should have let them increase their angle until the maximum and let them hold that, as opposed to just ignoring the command altogether. Like in cars, if you have REALLY good traction control, it will only step in at the VERY LAST minute to save you, when the car is SURE you've lost control. Or like a rev-limiter, to use another car analogy, it lets your run it up to the red, but it just doesn't let you go past that, instead of ignoring a command that *could* take you into the red. If an autonomous system is too overbearing, it very quickly becomes unsafe.
This, too, was excellent. Automation has been a huge plus for aviation, but even the best intentions can sometimes lead to unintended consequences, as it did here. But the net result is a software change that hadn't previously been considered necessary. I suspect there may still be other "gotchas" still in the Airbus software algorithms, but at least this one has, hopefully, been resolved.
Subscribed. I like your videos. I especially like that you narrate them. I’m not always looking at the screen but I can still can follow along. Also you do a good job of focusing in on relevant items.
There seems to be a common theme in computer-caused accidents: The computers are too reliant on single sensors and lack overall situational awareness. In this case, the computer(s) only considered alpha and not height above terrain, nor flight path angle, nor landing mode (short final). I find Airbus' adding the two stick inputs to be strange as well. This behavior is the same one that led to Air France 447 crashing as one pilot being full aft and the other being full forward prevented any attitude adjustment.
You say it’s a flaw but ultimately who designed the software…..a human. The software was modified soon after to disable high AOA protection below 200ft
@@AirspaceVideos Airbus has had numerous problems with their automation in the past. The big difference is they are honest about errors and fix them they do not hide behind corporate money and greed. And in Europe, you make a choice to risk passengers' lives and they die you will go to prison. Boeing just buys it's way through any criminal activity it commits with no one responsible ever to see justice. (My opinion)
Automation saved Air France 296. Pilots were much lower and slower than the planned altitude and speed, and were about to put the aircraft into a stall by pulling up. The fly by wire system prevented this, and thus the aircraft flew gently into the trees instead of falling from the air.
What is a better solution though for side-stick that aren’t connected? If the software was to follow the commands of the one which has move further from the central position then what if one is up and one is down?
the current solution isn't too bad in my opinion. want to take over? do it as airbus tells you and push the takeover button and you have sole control. It's just that people keep forgetting to do that, and that's hardly a design flaw I think.
@@AirspaceVideos , then Boeing shouldn’t be faulted for their assumption the pilots would flip the trim cutout switches. With a yoke instead of a sidestick at least you get the feedback from the other crews inputs.
Hmm that's comparing apples and oranges. The takeover button is clearly described, and so is the procedure for taking over. That was not the case for MCAS at the time.
It's been my understanding for many years that A-Prot was inhibited below 100 feet, not 200 feet as stated in the video. Great presentation and the clarification about pilot training between the US and Europe.
Quality video from a quality channel. First video I've seen of yours but subscribed immediately. Made me go on a binge of all your other content, great work!
I’m subscribed to several aviation tubes. A couple of months ago I subscribed to yours. Good work, keep it coming. You have a good narrative voice and decent understanding.
It seems like a better fix for this problem might be for the computer to simply raise the nose to the maximum safe angle of attack at that speed when the pilots pull back instead of completely cancelling their input. It also strikes me as being inherently a bad idea to have the computer add the inputs on both sides, especially with side sticks since it's not as easy to see if the other pilot is commanding any inputs. The altitude restriction will probably work for this particular situation but I think that the reason we still have two pilots in the cockpit is because humans can think outside of programming and therefore respond too strange situations in a way that computer cannot. Therefore, the computer should at least allow them to put in the maximum safe input at all altitudes and it should allows them to completely override the automation because the computer can only rely on instrumentation and, if something goes wrong with those instruments, it really takes a human pilot to deal with it.
I have never understood why Airbus choses to add sidestick commands when no take over button is pressed. It seems like a bad idea to me too, and it has been a factor in other accidents as well (air france 447 comes to mind)
The computer was programmed to raise the nose to the maximum angle of attack and hold it there with reasonable control inputs. In this case the inputs were excessive (both sticks hard back) and without computer intervention could have easily caused a stall. Of course with direct control the pilots would have almost immediately let off the sticks as the aircraft responded but the computer had no way of knowing that. With the stick inputs added, even then a slight misjudgement on either pilot's behalf could have brought them dangerously close to a stall.
When I moved to France and started working in France and flying a lot on what used to be called Air Inter I was blown away at how young the pilots were, even captains were often, it seemed, in their late, or even mid 20s, with an equally young copilot. I remember wondering whether, even if their skills were great, their judgement might still be a little "young." But their safety record is equal so I suppose my concern was misplaced.
I remember from another video somewhere talk about how a captain "takes control" by pressing that button. And holding it for like 15 seconds. So the computer would have given him control 10 seconds after they hit the ground...
This shows how automation faults will just become the new human factors. We write the software, therefore more automation will mean more accidents down to automation (this applies to cars, trains, planes…)
It is pure hubris on the part of airliner designers to think they can create a computer system that: A. Doesn't ever crash or otherwise malfunction; B. Can handle all circumstances and even override pilot inputs. It's ridiculous that pilots don't have the option to switch to basic stick-and-rudder flying techniques using basic instruments, for which all pilots should be well trained before even beginning training on planes such as these. Thanks for this video.
There is a way to force the pilots to have more control by forcing the plane into alternate law but that’s by deactivating certain systems which then do not make flying safe. The whole point is the pilot shouldn’t be doing something that causes the plane to step in. The few isolated events that have happened when it was actually safe (this, QF72, LH1829) were unforeseen events and nobody knew it could happen. Nobody knew the 737 was unsafe with the rudder PCU failure which caused 2 crashes. Nobody imagined an engine blowing up would cause total hydraulic loss on a DC-10
So is the computer too stupid to know that the plane is supposed to be stalled, or very close to it on landing? The last thing any pilot needs is to fight with a computer at a critical stage of flight...........
Indeed! My reaction when I first read that accident report was similar to yours. I was amazed that this could happen, since I know that the protection isn't active close to the ground. apparently it was, from 1988-2002...
@@glider1157 I've never flown a glider but you're right, the basic simplicity of it is quite intriguing........ I owned and flew a Piper Comanche (single engine, retractible gear) for a number of years, it had the older 'steam gauges', I enjoyed flying that plane way more than the big ones.
I don't understand why the computer is designed to add together stick inputs. That just sounds crazy to me. Shouldn't it be one stick or the other under all conditions? I could see weird pilot induced oscillations because of this as well if they both correct at the same time.
Sadly, it'd upset that copilot in a big way - did he even fly again ? Having heard that train drivers sometimes give the job up after someone commits suicide by jumping in front of the train.
Automation is amazing, it reacts faster than humans ever could, and it never tires, however, there needs to be several things to make it better, one is that it needs to be consistent and pilots need to be trained to its limitations; I've seen a lot of videos where automation doesn't tell you that it's been disengaged or that it doesn't know how to deal with a situation, the logic needs to be written so that it tells pilots hey I have this problem or this is not sustainable, in essence automation needs to be honest, and informative to the pilots as to its status, so the pilots are not caught by surprise if it fails all of a sudden
Garbage input will ways produce garbage output. Unfortunately, it is impossible to recreate all scenarios and situations where input can go wrong, from human to sensors.
Excellent description of what happened. Another example of system automation trying to be too smart. The main difference between Airbus and Boeing is that the computer is in control at Airbus and will only let the pilot do certain things while at Boeing, the pilot is always in control and the computer is a helper. I prefer a helper.
Oh dear me, are you serious? The indonesian, and ethiopian pilots who struggled with the Boeing 737 MAX computers would vehemently dissent - if only they could.
That's a complete lie considering mcas. Airbus planes have barely had such incidents you describe and if they have issues they hand control back to the pilots. Never a case with an airbus plane in which sole accident is down to the computer
Both planes had their issues with angle of attack protections. MCAS for Boeing, false Alpha Prot for Airbus. Refer to the XL Germany 888T accident, or the Lufthansa incident ar Bilbao.
@@AirspaceVideos 888T indeed was one but the computers deferred to direct law giving the pilots control without fbw till they geared up and even then the pilots failed to understand what was going on and fully stop their demonstration till the stall even struggling to add power which would have helped. In this case a very unlikely double aoa sensor failure happened and thus the 3rd working sensor was ignored as usually the logic to average two working ones failed as as essentially the logic has no way of comparing that third one is truly working aswell hence reverting to manual control.
Another dual input incident contributed to the accident. Side stick is a design flaw because there is no connection between the sticks so pilots cannot tell what the other pilot is inputting. Bad for CRM.
@@AirspaceVideos Dual input has been implicated in several high profile losses including AF447 and AirAsia 8501. Granted, there were system failures that caused the pilots to take extraordinary actions. In both accidents, the pilots were unaware the other was making control inputs so were unable to coordinate action.
@@encinobalboa neither pilot operated the priority take over push button which is there for that very reason. You tell the other pilot “I have control” push the button and carry out the recovery. And as you said both pilots did stupid things beforehand to end up in the situation they were left flailing at the controls. One crew failed to enact the correct procedures of unreliable airspeed and keeping the plane straight and level. The other disabled the FAC’s. Both situations do not mean loss of control
@@encinobalboa my point is there has never been a crash where the direct cause was because of both pilots using their sidesticks. There has always been another more prominent issue that caused the accidents and the lack of training meant that pilots forgot the button exists or ignored the DUAL INPUT warnings that the FWC blares out if you do so
The thing with automated system errors is it can almost always be traced back to a human error (or several) that lead to said failure. In this case whomever coded said system didn't account for this series of events & thus the accident occurred.
Computers are only as good as the programs written for them, and they may not deal appropriately with unusual or unique circumstances. But humans are not perfect, either.
Should they not have hit the TOGA right away? PIC does the approach, and co-pilot is ready to abort? Or in this case, the reverse. Wouldn't a TOGA disable the FCS natural instinct to prevent nose up?
Well they did exactly that, but since everything happened within the last ~6 seconds before impact, time was too short. When the captain tried to abort the approach, the sinkrate got too excessive to arrest and the engines did not have their time to respond (it usually takes 2-5 seconds for them to spin up to sufficient power on the A320). Unfortunately, the way the alpha protection was programmed, it did not cancel right away.
Things like this seem impossible to fix preemptively and it takes actual accidents beforehand. Automation is without a doubt better in consistency than a human and thus safer. Allowing a human to over ride it though defeats the purpose. That said bugs like this really put pilots in an impossible situation. Luckily they are rare and testing is critical.
Excellent video. Sometimes its the smallest things we miss. I wonder if the outcome might have been different if the CA had transferred control to the FO, then did the troubleshooting himself? He might have been more likely to find that breaker on the FO side.
@@AirspaceVideos Sorry! My comment was actually on the LOT gear-up landing. I had obviously followed the link in this video to that other one. But all are well-done!
So this kind of accident happened years ago involving some kind of "anti-stall feature"; it just showed us how bad boeing in executing the MCAS system, they didn't even introduce & explain the system to airlines & pilots.
Someone suggested to disconnect the alpha protection higher than 200 ft. I disagree, stall that close to the ground are unrecoverable and landing is the phase of flight when speed and AOA are closer to stall. I wonder why Airbus hasn’t used some other parameters or decisions logic to prevent this wrong input. How about an audio/visual warning to the pilots , to tell them what the computer wants to do and let them decide to override
I think additive control is there because when the captain goes one way and the first officer the other way, who wins out? We've seen that before but with bad CRM and putting in opposing controls, it seems the best thing to do is wait for the humans to figure out one of them needs to have sole control (either push the button or announce and acknowledge... and give up said controls.) Otherwise, the computer just adds +100 elevator with -100 elevator and gives you zero movement, as an example. But I'd they both put in +100 elevator, well it adds them together to and you get +200 elevator, out of its range. (I'm making up examples but you might see 100% directional inputs so let's just go maximum on it.) I'm sure there's a value assigned to each position of the stick or yoke ranging in positive to negative 100% as a way of conferring to the computer how much to move the control surfaces. Really it's about CRM. Someone take sole control as pilot flying. Switch off if you need to but communicate that.
OK, so someone wrote this software and didn't see that a) there is a problem when both pilots enter their inputs simultaneously and b) what stall at 5 or 20 m of altitude are we talking about. I mean, someone that programmed this had to think a bit about a variety of possibilities, not let the dead correct your bad code (in this case, luckily, no dead people). Some games have better writers.
Unfortunately, the aviation industry is often only willing to learn after something has happened. But this seems like a big oversight to me too. One could argue that the pilots mishandled the plane since they both used their sidesticks at the same time, which is not according to standard procedures. However, I think good software should prevent mishandling as much as possible.
@@AirspaceVideos moving up in the world! Lucky enough to be a neo? Vague as hell question, I know, but do they drive similar enough to the A320? Or would it be more like an A300/10? I seem to remember that it was either easy or was a common type rating between the 20/30, but I can't remember now. I also remember the avionics were just about identical, or are built on a common architecture. Would love to hear your thoughts on a narrow vs widebody! Thanks!
Well the basic philosophy behind both aircraft is very similar, but some systems are designed in a much more redundand way on the A330 due to ETOPS requirements etc. The cockpits feel very similar, and the flight feeling is similar, too. The A330 has a lot more mass tho and is gliding better, so descents have to be planned more accurately and with more leeway. Also, landing is a bit different, the flare starts at 50ft instead of 30ft on the A320, a thing that takes some getting used to. Also, during the conversion course, many new operational aspects have to be learned, such as ETOPS or in flight broadcasts over africa
Frightening to think that the people who program these computers can forget something that leads to a computer crashing a plane... Automation isn't to blame, the humans who didn't think about this possible scenario are!
One of the things that angers me is when people frame things as if "software error" and "human error" are two different categories. "We can eliminate human error by doing this in software and disallowing those pesky error-prone humans from overriding the computer.". Uhh the software is made by humans too. You just changed *which* human is in the driver's seat and moved the human error into the past, so it happens years ago at a desk rather than today in the sky.
In the eighties the F-16 which was largely fly by wire but in addition it had quadraredundant systems. Unfortunately, after about three dozen turned into lawn darts they had to rethink. I would submit that fly by wire is predominantly safe, but on that rare occasion a pilot has to be able to take command with manual control. On fighters if two wires short out at the stab actuator redundancy is worthless. So you need to be able to override with manual control.
If I'm not confused, the F16 is aerodynamically unsound. It requires FBW to fly at all. The instability also allows it to be highly maneuverable . It's a Fighter. The Airbus is not.
@@Dave_McKansas oh, my bad, I didn't realize you were uncertain that there is no difference in the principles of flight. My point had nothing to do with how many passengers are on an Airbus or what weapons armaments are on an F-16. My point was very concise. An aircraft, ANY aircraft, should allow the manual override failsafe because wires short out and computer hiccups happen.......... every day. BTW, thanks for your google insight. Fortunately, I didn't need google. I'm retired from boeing and my Air Force AFSC (Air Force Specialty Code) included F-15s, F-111s and F-16s. I was a Seven level with Red X orders. So you can put your google back in your holster.
it's similar in the light that both were AoA problems. However, this one was a very temporary one, while the 737max one was a.) persistent and b.) due to false readings and bad design choices (only 2 AoA probes)
Very clear explanation and summary of what went wrong, thank you. I am neutral on aircraft automation as I can see the up sides and the down. I am curious as to why the "Pitch Up" commands given by Capt. Sullenburger in an A320 ditching in the Hudson river, were also discarded just before water landing. If you have any insight I would be grateful thanks! 👍
Wow, I never heard about that. Just read a bit about it and it's rather complicated and has to do with airbus' dampening of the natural phugoid motion of an aircraft in flight. Also, the NTSB concluded that maybe Airbus should tell the pilots how exactly this works... "The NTSB concludes that training pilots that sidestick inputs may be attenuated when the airplane is in the alpha-protection mode would provide them with a better understanding of how entering the alpha-protection mode may affect the pitch response of the airplane. The NTSB recommends that the FAA require Airbus operators to expand the AOA-protection envelope limitations ground-school training to inform pilots about alpha-protection mode features while in normal law that can affect the pitch response of the airplane. "
This is a very interesting case. I just have one small criticism, where you mention old systems of mechanical control of flight surfaces. That is not really relevant. Fly-by-wire used in modern aircraft does not imply any particular level of automation, and certainly not that the computer should disregard the intentions of the pilots. This example goes to the heart of the different philosophies used by Airbus and Boeing in designing their flight control software. Traditionally, Boeing has given the pilot flying ultimate control over the aircraft. Airbus has given the computer more authority, then revised the software when incidents such as this one expose flaws. The B737 MAX was an exception to the rule, of course. Boeing badly botched it by taking control away from the pilot in a critical situation, and not recognizing the implications under fault conditions.
That programming seems maximally stupid. It just begs for these kinds of situations. It should not discard the input, but increase the AOA to the max. Just imagine you are hurling towards the ground at -10° of pitch and both of the pilots franticly pull their sticks in order to save the plane, it gets added and the computer evaluates the input would result of pitch over 30° and being over the max AOA for the plane and it continues to fly in high speed towards the ground killing everyone aboard. What kind of idiot wrote that code? "Hmm, I will discard this input if the input is too big instead of setting the input to the max allowed value, this is a good idea, people certainly do have to know what kind of input to a fly by wire system that provides no control feedback to the pilot and they don't increase the input when the plane is not doing anything after their input, that is a good idea!"
As far as I can see there is a major flaw in automation. By the time the pilot or driver realizes they need to intervene it's often going to be too late. Perhaps statistically its may be better to not allow proper intervention because overreaction and causing an issue is probably more common than the rare case where they know better than the computer. This issue has been highlighted to me watching videos of people using semi self-driving cars like the Tesla on Autopilot. In most of these videos everything will be going well then suddenly the driver has to make a correction. However, by the time the driver has reacted and made the correction the car has ended further off track than the would ever have been under manual control. Personally, I think seld driving cars or self-flying Aircraft will never come to the mainstream.
I think most companies consider any landing above 1.8 g, a hard landing report must be made... Would also be interesting if seeing the Capt didn’t take control and both pilots acted on the stick. Only a percentage of the combine input effects would have been ordered... I wonder.....🧐
It has been shown many times that when things go wrong the young pilots don’t have the experience to handle the situation. A younger pilot without experience would not have been able to do what Sully did.
The systems do the same, but Airbuses alpha protection is much more sophisticated and based on multiple sensors instead of just one like on MCAS. Also, this system had no part in this accident.
Yes, the tragedy is that , although automarion is noe essential and fails extremely rare, they seem so irrevokable and catastrophic! Maybe the engineers and software guys should ponder the "What ifs" a little more exhaustively.
Personally, I don't think that aircraft automation should be demonized, but I also think that aircraft demons are becoming a little bit too automated to my taste. Cheers!
I would say that both are inhibited so close to the ground. The two functions are very much interlinked so having one without the other makes no sense to me.
9:01 I'm not an expert, but in my humble opinion: if one pilot's inputs are deemed unacceptable to the computer, sure, it should throw it out. But if BOTH pilots input the same controls, perhaps this should initiate an automatic "takeover" command without anyone pressing the button. When fractions of a second matter, something like this could prevent a similar accident. Also, why it adds together the input from both side-sticks, is beyond me. That sounds like flawed programming, but again, I am no expert. If each input calls for 20 degrees pitch up, then pitch up 20 degrees; not 40. Or if one is 30, and one is 10, average them to 20. If both pilots are asking the same thing of the aircraft, but one pulls harder, and one softer - average them. Clearly the intention is the same; by what logic would you add them together?? "I'm not strong enough to pull a 4 oz. lever half an inch, we better both pull it and double the effect?" What? Maybe on a Boeing, or a small aircraft, but on an Airbus? What? I ask again, what is the logic here?
It's a very difficult problem and many people have many opinions on that. I'm sure the engineers at Airbus had a field day (or many) when they argued about how to design this logic...
I have, an absolutely fantastic presentation. It still applies today, and it's a big topic during many training sessions. Airbus even has the four golden rules on that topic: images.app.goo.gl/e6NoaVaFiViWE1iu6 But I don't think it applies here since the pilots were flying manually, and a hidden protection activated that the pilots could not have deactivated.
Issue with a design of any kind, and also an automation is .... have all the scenarios been taken into accout ?. Is this even possible ?. For me the answer is NO. The real customer, using the product every single day, constantly, with random combination of inputs and events, plus all the other customers using the same product .... can easily pop up a condition that has not code to deal with of to make thing worse, trigger an unrelated action. Is hard to test all the possible combinations and as much as an A320 can nail a land in zero visibility as soft as a feather, it can also misbehave. For instancwe, I dont trust the event of the paris airshow, I think this is more or less something like this accident, were some part of the system felt triggered but it wasnt need at all. Of course blaming pilot is cheaper. Thi is the reason why you wil NEVER be able to take pilots out of the cockpit. Computers cant weight all the given factor to take a disicion, like for example, ditching o landing in an open field or in a river like Sully. This is beyond any computer.
Well said, and I agree. Every program is only as smart as its creator(s). Only when true artificial intelligence is achieved can a true discussion about pilotless aircraft even start.
@@AirspaceVideos Ummmm I dont see artificial intelligence trying to deal with an expected escenario. This idea have been around for dacades. It could be good fo data trainining and analisis .... I didnt read much to be honest. Automation is all around us, not just a plane, but cars, lifts, trains. Short note: Long ago I worked for a casino machinery company. A machine from our competitors hanged badly during a play, it gave the player a ridiculous amount of money because it had crashed. My boss explained to me: This machines protect their transaction using CRC32. The chance that this validation fails is 1 in 4 billion. You will say ... WOW, exellent. No, its not. Because a machine can put 2000 games per day and all around the rooms are 200 to 300 units. And more on other casinos. It is a failure waiting to happen. His idea was more on the fail safe, that is, if everything crashes, the final routine that draws the display will reject out of range data and put another thing of just spaces. You CANT prevent a failure. You need to asume that something will go wrong anytime and by adding layers of control and protection, one over the other, and by keeping the basic idea of what means fail safe, this wont end into an accident. Aerospace itself is the prime example of perfection and fail safe ... still is hard so swallow that not other section of their systems flagged this condition as abnormal and let the pilots land the plane safely. Sames as Aifrance 447. Computer is getting garbage. Is giving overspeed and stall warning at the same time. Inside the code you can see this inconsistency, so just put a BIG MESSAGE on the screen to state that DATA is not longer RELIABLE and ask pilots to use alternate tools o measurements to fly the plane. Being Airbus state of the art in automation, this things puzzles me. That a tops notch plane keep popiing inconsistent alarms like a irresponsive cheap phone. Eventually this will drill and confuse pilots.
Well, this case is actually covered! If the speeds become unreliable or inconsistent, this calls for the unreliable speed procedure. But yes, I agree that we are far off from pilotless aircraft.
Don't blame the automation, blame the stupid software engineers. (I'm a software engineer and aviation enthusiast, I would never have made such a mistake if I'd written Airbus' software.) It is facepalm stuff. Does Airbus hire 3rd graders or something? *Facepalm*
if you're a sellout like almost every single content creators focusing on aviation here on UA-cam, i wouldn't recommend you to make a video about automation since it would piss me off and i would have to call you out, disrespectfully and expose your true motivations for creating such content...... but if you understand that the next step in aviation automation means AI and autonomous systems that will impact more than air transport safety ...... then go ahead an unbiased video on that particular topic would be awesome
Link to my other videos:
ua-cam.com/channels/IFpfSTM-3X3NpwVoh7VviQ.html
Would be interested in a video about automation dependency and the kinds of incidents/accidents we've seen as a result of it! Nice video by the way keep up the good work!
Thanks! I'll put it on my list and make a video about it at some point!
@@AirspaceVideos it's s delicate subject to cover because it's easy to report on the bias rather than the actual effects. What I mean is, that if automation saves you it's not as news worthy as when automation fails.
If you cover it I think you'd need to compare cases that are completely connected. So here the alpha prot caused a crash but you'd need to give examples where alpha prot prevented a crash and you'd need to give numbers and statistics.
It definitely is an interesting topic but it's super delicate as well and needs a huge amount of research and i doubt that it would fit into a 10-15minute time frame tbh.
I would like to see a video about the technological advances in aviation. Pros and cons.
Full automation is the only way to go. This incident then wouldn't have happened - because the automation would be so thoroughly checked, double checked, triple checked and quadruple checked by independent inspectors within the company and likely one from another company - like an insurance company.
This is something that has always worried me. Having lived in Wellington New Zealand I, as a passenger have been very aware of the extreme and often fairly unconventional techniques required to get planes down on the ground in these sorts of conditions. These techniques usually involve keeping high levels of trust on right up to touchdown.
I am really not sure how they do all this stuff in a modern plane. I would like to hear your thoughts on automation and iit's a substitute for actually flying the plane manually.
I think automation is great, overall. 99.9% of the time, it works just as intended, and that's great. I wouldn't want to fly a transatlantic flight fully by hand, that would be insanely tiresome (and boring). There are also times, where automation can help greatly. For example, when I fly an approach in very gusty weather, I like to delegate the task of maintaining airspeed to the autothrust. It does a great job of keeping the appropriate speed and I have more mental capacity on getting the plane's trajectory just right. However, this does not free me from controlling if the autothrust does it correctly, and I must always know what it does. If I don't, I switch it off.
However, no system is infallible, and like that, accidents like this one happen. It's very unfortunate, but also very very rare.
@@AirspaceVideos I like your answer, the thing I pick up from it is that you are using the automation to help you fly. However, you are not letting the automation control you.
exactly! These are two of the golden rules of automation usage - know what it does at all times and take control if it does things that you don't like.
I like to see it as cruise control in a car. If you turn it on, it takes over and maintains speed so you can focus on other things, but at no point does it remove the ability to control the car from you. If you tap the brakes, it disconnects, and if you accelerate manually, the cruise input and your input are effectively added to each other. If you hit autopilot disconnect, same thing happens, the autopilot disconnects immediately and you are now flying the plane.
I think automation can be a great thing, but one of automation's biggest dangers is the designers thinking they know better than the pilot.
I don't know this system well, but I'd say that it should have let them increase their angle until the maximum and let them hold that, as opposed to just ignoring the command altogether.
Like in cars, if you have REALLY good traction control, it will only step in at the VERY LAST minute to save you, when the car is SURE you've lost control. Or like a rev-limiter, to use another car analogy, it lets your run it up to the red, but it just doesn't let you go past that, instead of ignoring a command that *could* take you into the red. If an autonomous system is too overbearing, it very quickly becomes unsafe.
good analogy! I think getting automation "just right" is the hard part. not too sloppy, not too overbearing...
This, too, was excellent. Automation has been a huge plus for aviation, but even the best intentions can sometimes lead to unintended consequences, as it did here. But the net result is a software change that hadn't previously been considered necessary. I suspect there may still be other "gotchas" still in the Airbus software algorithms, but at least this one has, hopefully, been resolved.
Subscribed. I like your videos. I especially like that you narrate them. I’m not always looking at the screen but I can still can follow along. Also you do a good job of focusing in on relevant items.
Thank you :) That's what I try to do.
There seems to be a common theme in computer-caused accidents: The computers are too reliant on single sensors and lack overall situational awareness. In this case, the computer(s) only considered alpha and not height above terrain, nor flight path angle, nor landing mode (short final). I find Airbus' adding the two stick inputs to be strange as well. This behavior is the same one that led to Air France 447 crashing as one pilot being full aft and the other being full forward prevented any attitude adjustment.
You say it’s a flaw but ultimately who designed the software…..a human. The software was modified soon after to disable high AOA protection below 200ft
I dont work in aviation sector but these videos are so easily understandable. You are more than a Professional!!!!
I imagine that had their been fatalities this would have been Airbus's MCAS moment.
Airbus had many MCAS moments before actually. check out XL airways crash or lufthansa ar bilbao!
@@AirspaceVideos Airbus has had numerous problems with their automation in the past. The big difference is they are honest about errors and fix them they do not hide behind corporate money and greed. And in Europe, you make a choice to risk passengers' lives and they die you will go to prison. Boeing just
buys it's way through any criminal activity it commits with no one responsible ever to see justice. (My opinion)
@@kikufutaba1194 , 4 BILLION dollar fine for Airbus Bribery, hardly squeaky clean....how’s that for corporate money and greed? Honesty? SMH
Agreed
The A320 protections can sometimes worsen the situations, but in general it works very well, not the case here.
Automation saved Air France 296. Pilots were much lower and slower than the planned altitude and speed, and were about to put the aircraft into a stall by pulling up. The fly by wire system prevented this, and thus the aircraft flew gently into the trees instead of falling from the air.
Additive inputs from dual controls is a cockamamie design, in my opinion.
What is a better solution though for side-stick that aren’t connected? If the software was to follow the commands of the one which has move further from the central position then what if one is up and one is down?
the current solution isn't too bad in my opinion. want to take over? do it as airbus tells you and push the takeover button and you have sole control. It's just that people keep forgetting to do that, and that's hardly a design flaw I think.
@@AirspaceVideos , then Boeing shouldn’t be faulted for their assumption the pilots would flip the trim cutout switches. With a yoke instead of a sidestick at least you get the feedback from the other crews inputs.
Hmm that's comparing apples and oranges. The takeover button is clearly described, and so is the procedure for taking over. That was not the case for MCAS at the time.
Really glad I have found your channel! All your videos are so interesting and very well made! :D
thanks a lot :) have fun watching!
It's been my understanding for many years that A-Prot was inhibited below 100 feet, not 200 feet as stated in the video. Great presentation and the clarification about pilot training between the US and Europe.
That may indeed be the case! The A-Prot has been changed many times over the years.
Thanks a lot for watching :)
Quality video from a quality channel. First video I've seen of yours but subscribed immediately. Made me go on a binge of all your other content, great work!
Wow, thank you! Many more to come :)
I’m subscribed to several aviation tubes. A couple of months ago I subscribed to yours. Good work, keep it coming. You have a good narrative voice and decent understanding.
I immediately thought, "Oh no, they triggered the alpha floor and alpha max protection." Boom. Nose down.
It seems like a better fix for this problem might be for the computer to simply raise the nose to the maximum safe angle of attack at that speed when the pilots pull back instead of completely cancelling their input. It also strikes me as being inherently a bad idea to have the computer add the inputs on both sides, especially with side sticks since it's not as easy to see if the other pilot is commanding any inputs. The altitude restriction will probably work for this particular situation but I think that the reason we still have two pilots in the cockpit is because humans can think outside of programming and therefore respond too strange situations in a way that computer cannot. Therefore, the computer should at least allow them to put in the maximum safe input at all altitudes and it should allows them to completely override the automation because the computer can only rely on instrumentation and, if something goes wrong with those instruments, it really takes a human pilot to deal with it.
I have never understood why Airbus choses to add sidestick commands when no take over button is pressed. It seems like a bad idea to me too, and it has been a factor in other accidents as well (air france 447 comes to mind)
The computer was programmed to raise the nose to the maximum angle of attack and hold it there with reasonable control inputs. In this case the inputs were excessive (both sticks hard back) and without computer intervention could have easily caused a stall. Of course with direct control the pilots would have almost immediately let off the sticks as the aircraft responded but the computer had no way of knowing that. With the stick inputs added, even then a slight misjudgement on either pilot's behalf could have brought them dangerously close to a stall.
When I moved to France and started working in France and flying a lot on what used to be called Air Inter I was blown away at how young the pilots were, even captains were often, it seemed, in their late, or even mid 20s, with an equally young copilot. I remember wondering whether, even if their skills were great, their judgement might still be a little "young." But their safety record is equal so I suppose my concern was misplaced.
Thank you! I really enjoy these vids. Keep them coming 👍
Glad you like them! Many more to come :)
I remember from another video somewhere talk about how a captain "takes control" by pressing that button. And holding it for like 15 seconds. So the computer would have given him control 10 seconds after they hit the ground...
This shows how automation faults will just become the new human factors. We write the software, therefore more automation will mean more accidents down to automation (this applies to cars, trains, planes…)
It is pure hubris on the part of airliner designers to think they can create a computer system that: A. Doesn't ever crash or otherwise malfunction; B. Can handle all circumstances and even override pilot inputs. It's ridiculous that pilots don't have the option to switch to basic stick-and-rudder flying techniques using basic instruments, for which all pilots should be well trained before even beginning training on planes such as these. Thanks for this video.
There is a way to force the pilots to have more control by forcing the plane into alternate law but that’s by deactivating certain systems which then do not make flying safe. The whole point is the pilot shouldn’t be doing something that causes the plane to step in. The few isolated events that have happened when it was actually safe (this, QF72, LH1829) were unforeseen events and nobody knew it could happen. Nobody knew the 737 was unsafe with the rudder PCU failure which caused 2 crashes. Nobody imagined an engine blowing up would cause total hydraulic loss on a DC-10
Thanks for sharing. Your channel looks promising. Will certainly follow it.
Thanks! Looking forward to growing it :)
A nice big red button with emergency manual override written on it would be a great feature ..
So is the computer too stupid to know that the plane is supposed to be stalled, or very close to it on landing?
The last thing any pilot needs is to fight with a computer at a critical stage of flight...........
Indeed! My reaction when I first read that accident report was similar to yours. I was amazed that this could happen, since I know that the protection isn't active close to the ground. apparently it was, from 1988-2002...
Glad not having such gadgets in the gliders I fly....
@@glider1157 I've never flown a glider but you're right, the basic simplicity of it is quite intriguing........
I owned and flew a Piper Comanche (single engine, retractible gear) for a number of years, it had the older 'steam gauges', I enjoyed flying that plane way more than the big ones.
I don't understand why the computer is designed to add together stick inputs. That just sounds crazy to me. Shouldn't it be one stick or the other under all conditions? I could see weird pilot induced oscillations because of this as well if they both correct at the same time.
yeah, it's a debatable design choice
Sadly, it'd upset that copilot in a big way - did he even fly again ?
Having heard that train drivers sometimes give the job up after someone commits suicide by jumping in front of the train.
no info on that, unfortunately
Automation is amazing, it reacts faster than humans ever could, and it never tires, however, there needs to be several things to make it better, one is that it needs to be consistent and pilots need to be trained to its limitations; I've seen a lot of videos where automation doesn't tell you that it's been disengaged or that it doesn't know how to deal with a situation, the logic needs to be written so that it tells pilots hey I have this problem or this is not sustainable, in essence automation needs to be honest, and informative to the pilots as to its status, so the pilots are not caught by surprise if it fails all of a sudden
Garbage input will ways produce garbage output. Unfortunately, it is impossible to recreate all scenarios and situations where input can go wrong, from human to sensors.
Excellent description of what happened. Another example of system automation trying to be too smart. The main difference between Airbus and Boeing is that the computer is in control at Airbus and will only let the pilot do certain things while at Boeing, the pilot is always in control and the computer is a helper. I prefer a helper.
Oh dear me, are you serious? The indonesian, and ethiopian pilots who struggled with the Boeing 737 MAX computers would vehemently dissent - if only they could.
That's a complete lie considering mcas. Airbus planes have barely had such incidents you describe and if they have issues they hand control back to the pilots. Never a case with an airbus plane in which sole accident is down to the computer
Both planes had their issues with angle of attack protections. MCAS for Boeing, false Alpha Prot for Airbus. Refer to the XL
Germany 888T accident, or the Lufthansa incident ar Bilbao.
@@AirspaceVideos 888T indeed was one but the computers deferred to direct law giving the pilots control without fbw till they geared up and even then the pilots failed to understand what was going on and fully stop their demonstration till the stall even struggling to add power which would have helped. In this case a very unlikely double aoa sensor failure happened and thus the 3rd working sensor was ignored as usually the logic to average two working ones failed as as essentially the logic has no way of comparing that third one is truly working aswell hence reverting to manual control.
Apparently you have forgotten the 737 max. Not to mention Turkish Airlines crash in Amsterdam.
Very interested in how automation works and is coordinated with expected and unexpected human behavior.
Another dual input incident contributed to the accident. Side stick is a design flaw because there is no connection between the sticks so pilots cannot tell what the other pilot is inputting. Bad for CRM.
I wouldn't call the sidesticks a design flaw. If they are handled correctly, they work perfectly fine. I just think the dual input logic is terrible!
@@AirspaceVideos Dual input has been implicated in several high profile losses including AF447 and AirAsia 8501. Granted, there were system failures that caused the pilots to take extraordinary actions. In both accidents, the pilots were unaware the other was making control inputs so were unable to coordinate action.
@@encinobalboa neither pilot operated the priority take over push button which is there for that very reason. You tell the other pilot “I have control” push the button and carry out the recovery. And as you said both pilots did stupid things beforehand to end up in the situation they were left flailing at the controls. One crew failed to enact the correct procedures of unreliable airspeed and keeping the plane straight and level. The other disabled the FAC’s. Both situations do not mean loss of control
@@tomstravels520 Your line of reasoning suggests simpler controls without take over button would have served them better in an emergency situation.
@@encinobalboa my point is there has never been a crash where the direct cause was because of both pilots using their sidesticks. There has always been another more prominent issue that caused the accidents and the lack of training meant that pilots forgot the button exists or ignored the DUAL INPUT warnings that the FWC blares out if you do so
Seems like the alpha protection wasn’t working so well on Air France flight 447 ?
Exactly, it didn't! It couldn't since the speed probes had iced over and the flight conputers changed to a backup control law.
Alpha floor active until 100 Feet , by the way good job keep going 👍👍
The thing with automated system errors is it can almost always be traced back to a human error (or several) that lead to said failure. In this case whomever coded said system didn't account for this series of events & thus the accident occurred.
Computers are only as good as the programs written for them, and they may not deal appropriately with unusual or unique circumstances. But humans are not perfect, either.
Should they not have hit the TOGA right away? PIC does the approach, and co-pilot is ready to abort? Or in this case, the reverse. Wouldn't a TOGA disable the FCS natural instinct to prevent nose up?
Well they did exactly that, but since everything happened within the last ~6 seconds before impact, time was too short. When the captain tried to abort the approach, the sinkrate got too excessive to arrest and the engines did not have their time to respond (it usually takes 2-5 seconds for them to spin up to sufficient power on the A320). Unfortunately, the way the alpha protection was programmed, it did not cancel right away.
Things like this seem impossible to fix preemptively and it takes actual accidents beforehand. Automation is without a doubt better in consistency than a human and thus safer. Allowing a human to over ride it though defeats the purpose. That said bugs like this really put pilots in an impossible situation. Luckily they are rare and testing is critical.
4:45 that is a dangerous concentration of pencils o.o
Excellent video. Sometimes its the smallest things we miss. I wonder if the outcome might have been different if the CA had transferred control to the FO, then did the troubleshooting himself? He might have been more likely to find that breaker on the FO side.
Thank you! Well yes, maybe.
@@AirspaceVideos Sorry! My comment was actually on the LOT gear-up landing. I had obviously followed the link in this video to that other one. But all are well-done!
So this kind of accident happened years ago involving some kind of "anti-stall feature"; it just showed us how bad boeing in executing the MCAS system, they didn't even introduce & explain the system to airlines & pilots.
So you think the Airbus pilots on this flight knew? I think not, there’s a broken airplane ....
@@stevegiboney4493 What I mean is that Boeing in designing the MCAS system should have learned from this accident.
Nice to know you’re an airline pilot. What planes do you fly?
A320 previously, now A330
the animation is sooooooooooo good
What an amazing vid. Subbed!
Thanks! Welcome to the channel!
I feel very bad for that first officer
A very good video. Well explained. Please do a video on automation.
Thanks, I will!
Yo you won a new subscriber congratulations
thanks!
Disable the safety feature at 200ft?? A little low you think? maybe 500-700ft would be more reasonable I think..
Probably the predicting part of the protection was like too much
Someone suggested to disconnect the alpha protection higher than 200 ft. I disagree, stall that close to the ground are unrecoverable and landing is the phase of flight when speed and AOA are closer to stall. I wonder why Airbus hasn’t used some other parameters or decisions logic to prevent this wrong input. How about an audio/visual warning to the pilots , to tell them what the computer wants to do and let them decide to override
From what I understand the high AOA protection is disabled below 200ft but the Alpha Floor still operates until 100ft
I think additive control is there because when the captain goes one way and the first officer the other way, who wins out? We've seen that before but with bad CRM and putting in opposing controls, it seems the best thing to do is wait for the humans to figure out one of them needs to have sole control (either push the button or announce and acknowledge... and give up said controls.) Otherwise, the computer just adds +100 elevator with -100 elevator and gives you zero movement, as an example. But I'd they both put in +100 elevator, well it adds them together to and you get +200 elevator, out of its range. (I'm making up examples but you might see 100% directional inputs so let's just go maximum on it.)
I'm sure there's a value assigned to each position of the stick or yoke ranging in positive to negative 100% as a way of conferring to the computer how much to move the control surfaces.
Really it's about CRM. Someone take sole control as pilot flying. Switch off if you need to but communicate that.
OK, so someone wrote this software and didn't see that a) there is a problem when both pilots enter their inputs simultaneously and b) what stall at 5 or 20 m of altitude are we talking about. I mean, someone that programmed this had to think a bit about a variety of possibilities, not let the dead correct your bad code (in this case, luckily, no dead people). Some games have better writers.
Unfortunately, the aviation industry is often only willing to learn after something has happened. But this seems like a big oversight to me too. One could argue that the pilots mishandled the plane since they both used their sidesticks at the same time, which is not according to standard procedures. However, I think good software should prevent mishandling as much as possible.
May I ask what aircraft you normally fly please mate? Also, are you a Boeing or Airbus enthusiast?
As far as I know he's an A320 pilot.
oh sorry, missed your comment. I flew the A320 for 5 years, now I'm on the A330.
@@AirspaceVideos moving up in the world! Lucky enough to be a neo? Vague as hell question, I know, but do they drive similar enough to the A320? Or would it be more like an A300/10? I seem to remember that it was either easy or was a common type rating between the 20/30, but I can't remember now.
I also remember the avionics were just about identical, or are built on a common architecture. Would love to hear your thoughts on a narrow vs widebody! Thanks!
Well the basic philosophy behind both aircraft is very similar, but some systems are designed in a much more redundand way on the A330 due to ETOPS requirements etc. The cockpits feel very similar, and the flight feeling is similar, too. The A330 has a lot more mass tho and is gliding better, so descents have to be planned more accurately and with more leeway. Also, landing is a bit different, the flare starts at 50ft instead of 30ft on the A320, a thing that takes some getting used to. Also, during the conversion course, many new operational aspects have to be learned, such as ETOPS or in flight broadcasts over africa
Exactly like first flight of this plane and crash at airshow? same thing happened
Frightening to think that the people who program these computers can forget something that leads to a computer crashing a plane... Automation isn't to blame, the humans who didn't think about this possible scenario are!
Exactly, a program is always only as good as its creators!
One of the things that angers me is when people frame things as if "software error" and "human error" are two different categories. "We can eliminate human error by doing this in software and disallowing those pesky error-prone humans from overriding the computer.". Uhh the software is made by humans too. You just changed *which* human is in the driver's seat and moved the human error into the past, so it happens years ago at a desk rather than today in the sky.
In the eighties the F-16 which was largely fly by wire but in addition it had quadraredundant systems. Unfortunately, after about three dozen turned into lawn darts they had to rethink. I would submit that fly by wire is predominantly safe, but on that rare occasion a pilot has to be able to take command with manual control. On fighters if two wires short out at the stab actuator redundancy is worthless. So you need to be able to override with manual control.
If I'm not confused, the F16 is aerodynamically unsound. It requires FBW to fly at all. The instability also allows it to be highly maneuverable . It's a Fighter.
The Airbus is not.
@@Dave_McKansas oh, my bad, I didn't realize you were uncertain that there is no difference in the principles of flight. My point had nothing to do with how many passengers are on an Airbus or what weapons armaments are on an F-16. My point was very concise. An aircraft, ANY aircraft, should allow the manual override failsafe because wires short out and computer hiccups happen.......... every day. BTW, thanks for your google insight. Fortunately, I didn't need google. I'm retired from boeing and my Air Force AFSC (Air Force Specialty Code) included F-15s, F-111s and F-16s. I was a Seven level with Red X orders. So you can put your google back in your holster.
How’s this problem different from the 737 Max problem?
it's similar in the light that both were AoA problems. However, this one was a very temporary one, while the 737max one was a.) persistent and b.) due to false readings and bad design choices (only 2 AoA probes)
1:01 Ooh ... lovely A340!
Right? The A340-600 is a sight to see! Pretty rare these days though.
I flew a plane once. It was the best 10 seconds of my life.
Writing from heavens ?
Did the young pilot keep flying?
No info on that available, but I don't see why not!
Very clear explanation and summary of what went wrong, thank you. I am neutral on aircraft automation as I can see the up sides and the down. I am curious as to why the "Pitch Up" commands given by Capt. Sullenburger in an A320 ditching in the Hudson river, were also discarded just before water landing. If you have any insight I would be grateful thanks! 👍
Wow, I never heard about that. Just read a bit about it and it's rather complicated and has to do with airbus' dampening of the natural phugoid motion of an aircraft in flight.
Also, the NTSB concluded that maybe Airbus should tell the pilots how exactly this works...
"The NTSB concludes that training pilots that sidestick inputs may be attenuated when the airplane is in the alpha-protection mode would provide them with a better understanding of how entering the alpha-protection mode may affect the pitch response of the airplane. The NTSB recommends that the FAA require Airbus operators to expand the AOA-protection envelope limitations ground-school training to inform pilots about alpha-protection mode features while in normal law that can affect the pitch response of the airplane. "
The fiat aircraft crash in Turin 4th May 1949 might be a useful topic for a vid no automation then
This is a very interesting case. I just have one small criticism, where you mention old systems of mechanical control of flight surfaces. That is not really relevant. Fly-by-wire used in modern aircraft does not imply any particular level of automation, and certainly not that the computer should disregard the intentions of the pilots. This example goes to the heart of the different philosophies used by Airbus and Boeing in designing their flight control software. Traditionally, Boeing has given the pilot flying ultimate control over the aircraft. Airbus has given the computer more authority, then revised the software when incidents such as this one expose flaws.
The B737 MAX was an exception to the rule, of course. Boeing badly botched it by taking control away from the pilot in a critical situation, and not recognizing the implications under fault conditions.
That programming seems maximally stupid. It just begs for these kinds of situations. It should not discard the input, but increase the AOA to the max. Just imagine you are hurling towards the ground at -10° of pitch and both of the pilots franticly pull their sticks in order to save the plane, it gets added and the computer evaluates the input would result of pitch over 30° and being over the max AOA for the plane and it continues to fly in high speed towards the ground killing everyone aboard. What kind of idiot wrote that code? "Hmm, I will discard this input if the input is too big instead of setting the input to the max allowed value, this is a good idea, people certainly do have to know what kind of input to a fly by wire system that provides no control feedback to the pilot and they don't increase the input when the plane is not doing anything after their input, that is a good idea!"
Computers try to prevent certain situations, but create other situations by doing so.
Hello Can you do a video on a topic? It has to with coffee and two Airbusses that are NEW?
do you mean the A350 coffee engine shutdown?
@@AirspaceVideos YES!
I might do that!
@@AirspaceVideos Awsome !
great video
merci :)
As far as I can see there is a major flaw in automation. By the time the pilot or driver realizes they need to intervene it's often going to be too late. Perhaps statistically its may be better to not allow proper intervention because overreaction and causing an issue is probably more common than the rare case where they know better than the computer.
This issue has been highlighted to me watching videos of people using semi self-driving cars like the Tesla on Autopilot. In most of these videos everything will be going well then suddenly the driver has to make a correction. However, by the time the driver has reacted and made the correction the car has ended further off track than the would ever have been under manual control.
Personally, I think seld driving cars or self-flying Aircraft will never come to the mainstream.
I think most companies consider any landing above 1.8 g, a hard landing report must be made...
Would also be interesting if seeing the Capt didn’t take control and both pilots acted on the stick. Only a percentage of the combine input effects would have been ordered... I wonder.....🧐
It has been shown many times that when things go wrong the young pilots don’t have the experience to handle the situation. A younger pilot without experience would not have been able to do what Sully did.
Age doesn't matter, experience does. That's why there is always an experienced individual on the flight deck, the captain.
The only way to become an experienced pilot is to start out as an inexperienced one. You can't get rid of inexperienced pilots.
true!
Pure speculation
Looks like MCAS in Boeing, Alpha Protection in Airbus...? Different purpose but same result
The systems do the same, but Airbuses alpha protection is much more sophisticated and based on multiple sensors instead of just one like on MCAS. Also, this system had no part in this accident.
Quite hard to steal from your own company isn't it
Skip to 2:03 to start
How does this relate to AF 447? Is the Alpha protection system fundamentally different in the A-330?
It works exactly the same on the A320 and the A330. The two cases are aoa-related, but entirely different (one in cruise, the other on landing).
in AF447, the pilots mishandled an approach to stall, in this case the alpha prot reacted unexpectedly
Yes, the tragedy is that , although automarion is noe essential and fails extremely rare, they seem so irrevokable and catastrophic! Maybe the engineers and software guys should ponder the "What ifs" a little more exhaustively.
Essentially. Had the flown it as Airbus FCOM and FCTM require this wouldn’t have happened.
Maybe give the autopilot a voice so it could participate in CRM...?
Thank you 👌
you're welcome!
Good video but please spell check your captioning particularly the word February. 🙏🏻👍🏻🌞
thank you :) note that captions are done automatically by youtube.
Personally, I don't think that aircraft automation should be demonized, but I also think that aircraft demons are becoming a little bit too automated to my taste. Cheers!
😄 damn those demons...
Alpha prot is inhibited but alpha floor is not inhibited right?
in what case?
@@AirspaceVideos at 200ft agl
I would say that both are inhibited so close to the ground. The two functions are very much interlinked so having one without the other makes no sense to me.
@@AirspaceVideos so does that mean it's possible to stall an airbus below 200ft?
Exactly, theoretically. But if you made it to 200ft while not being in a stall, most likely the last 200ft will go well too ;)
9:01 I'm not an expert, but in my humble opinion: if one pilot's inputs are deemed unacceptable to the computer, sure, it should throw it out. But if BOTH pilots input the same controls, perhaps this should initiate an automatic "takeover" command without anyone pressing the button. When fractions of a second matter, something like this could prevent a similar accident.
Also, why it adds together the input from both side-sticks, is beyond me. That sounds like flawed programming, but again, I am no expert. If each input calls for 20 degrees pitch up, then pitch up 20 degrees; not 40. Or if one is 30, and one is 10, average them to 20. If both pilots are asking the same thing of the aircraft, but one pulls harder, and one softer - average them. Clearly the intention is the same; by what logic would you add them together?? "I'm not strong enough to pull a 4 oz. lever half an inch, we better both pull it and double the effect?" What? Maybe on a Boeing, or a small aircraft, but on an Airbus? What? I ask again, what is the logic here?
It's a very difficult problem and many people have many opinions on that. I'm sure the engineers at Airbus had a field day (or many) when they argued about how to design this logic...
Hope Airbus was held liable.
No. It's absolutely proper to demonize automation.
Pilots should not be asking a computer for permission to fly the plane.
Have you seen 'Children of the Magenta Line'?
I have, an absolutely fantastic presentation. It still applies today, and it's a big topic during many training sessions. Airbus even has the four golden rules on that topic:
images.app.goo.gl/e6NoaVaFiViWE1iu6
But I don't think it applies here since the pilots were flying manually, and a hidden protection activated that the pilots could not have deactivated.
@@AirspaceVideos Not saying it does here, but it discuses a different but important part of the whole flight automation issue.
Oh yes, indeed!
@@AirspaceVideos , the concept of both pilots control inputs adding together is a definite flaw in my opinion.
Did the trainee pilot recover his confidence and go onto have a career as a pilot ?
I don't know, but I hope so!
at least it wasn't MCAS
😑
‘Scarebus’
Boing Boing
Too many ads. Can’t listen
The only to be blamed is the computerised, automated and French plastic robot aircraft!
last time I checked, the A320 was pretty much entirely made of metal 😛
Issue with a design of any kind, and also an automation is .... have all the scenarios been taken into accout ?. Is this even possible ?. For me the answer is NO. The real customer, using the product every single day, constantly, with random combination of inputs and events, plus all the other customers using the same product .... can easily pop up a condition that has not code to deal with of to make thing worse, trigger an unrelated action. Is hard to test all the possible combinations and as much as an A320 can nail a land in zero visibility as soft as a feather, it can also misbehave. For instancwe, I dont trust the event of the paris airshow, I think this is more or less something like this accident, were some part of the system felt triggered but it wasnt need at all. Of course blaming pilot is cheaper. Thi is the reason why you wil NEVER be able to take pilots out of the cockpit. Computers cant weight all the given factor to take a disicion, like for example, ditching o landing in an open field or in a river like Sully. This is beyond any computer.
Well said, and I agree. Every program is only as smart as its creator(s). Only when true artificial intelligence is achieved can a true discussion about pilotless aircraft even start.
@@AirspaceVideos Ummmm I dont see artificial intelligence trying to deal with an expected escenario. This idea have been around for dacades. It could be good fo data trainining and analisis .... I didnt read much to be honest. Automation is all around us, not just a plane, but cars, lifts, trains. Short note: Long ago I worked for a casino machinery company. A machine from our competitors hanged badly during a play, it gave the player a ridiculous amount of money because it had crashed. My boss explained to me: This machines protect their transaction using CRC32. The chance that this validation fails is 1 in 4 billion. You will say ... WOW, exellent. No, its not. Because a machine can put 2000 games per day and all around the rooms are 200 to 300 units. And more on other casinos. It is a failure waiting to happen.
His idea was more on the fail safe, that is, if everything crashes, the final routine that draws the display will reject out of range data and put another thing of just spaces. You CANT prevent a failure. You need to asume that something will go wrong anytime and by adding layers of control and protection, one over the other, and by keeping the basic idea of what means fail safe, this wont end into an accident. Aerospace itself is the prime example of perfection and fail safe ... still is hard so swallow that not other section of their systems flagged this condition as abnormal and let the pilots land the plane safely. Sames as Aifrance 447. Computer is getting garbage. Is giving overspeed and stall warning at the same time. Inside the code you can see this inconsistency, so just put a BIG MESSAGE on the screen to state that DATA is not longer RELIABLE and ask pilots to use alternate tools o measurements to fly the plane. Being Airbus state of the art in automation, this things puzzles me. That a tops notch plane keep popiing inconsistent alarms like a irresponsive cheap phone. Eventually this will drill and confuse pilots.
Well, this case is actually covered! If the speeds become unreliable or inconsistent, this calls for the unreliable speed procedure.
But yes, I agree that we are far off from pilotless aircraft.
I hope that Tesla will never try to build an airplane...
I'D LIKE THAT PLANE
Don't blame the automation, blame the stupid software engineers. (I'm a software engineer and aviation enthusiast, I would never have made such a mistake if I'd written Airbus' software.) It is facepalm stuff. Does Airbus hire 3rd graders or something? *Facepalm*
if you're a sellout like almost every single content creators focusing on aviation here on UA-cam, i wouldn't recommend you to make a video about automation since it would piss me off and i would have to call you out, disrespectfully and expose your true motivations for creating such content...... but if you understand that the next step in aviation automation means AI and autonomous systems that will impact more than air transport safety ...... then go ahead an unbiased video on that particular topic would be awesome
Mmmmmmmm......... I wonder if aircraft automation could ever cause an aircraft to crash?🤔🤔
Boeing 737 Max anyone?🤔🤔
Soul plane...the crew get their license from Microsoft flight sim.....FO where is the reset button? CATAIN....WHAT? OMG.
I can't even decipher what you are trying to say, sorry.
They don’t call them the scarebus for nothing 😆
cough cough, 737 and MCAS, you know 🙃 no plane's perfect!
@@AirspaceVideos Boeing learned from airbus.
Only cowards call them scarebus
@@AirspaceVideos Only if airbus fan boys would understand that. As a person who likes both I'm kinda hearing 'the flying coffin' stuff
Sir can you tell what planes do you fly
A320 and now A330
Nose first landing
Ounch
Nice video
thanks!