not sure it's legit for airbus to be declaring a win for their automation. sounds like the copilot dogpaddled back to the cockpit through freefall and batted at his controls from the ceiling until the computer got confused and resumed level flight.
If any safety technology deserves credit here it's seatbelts. From the sound of it, nobody who was belted in was injured.
The computer worked to prevent the aircraft from entering an overspeed condition by pitching up and attempting to recover from the dive even before that dual stick input was an issue. It was intervening 13 seconds into the incident, as the aircraft threatened to exceed safety tolerances. The dual stick input created another issue after that intervention began.
The automation very possibly saved lives here. The overspeed protection is good because had that not happened, there's a real possibility that significant sections of the airplane would have sheared off, or the airplane could have rapidly drifted further out of control as it accelerated quickly through 365 knots indicated. Big, heavy, thrusting things gain airspeed rapidly when you pitch them down 18 degrees at cruise power, and the margin of recoverability is very minimal. There are arguments that humans are unable to prevent some scenarios in that situation because by the time higher-order thinking on decisions has happened, it's too late. See, for example, the surprisingly lengthy duration of time between the incident and setting for idle, which probably should have been among the first things done in response. Brains are slow when things happen fast, and a symphony of alarms and whoops (undoubtedly in progress) does not help.
A Boeing would not have intervened in the situation at all and would have allowed the aircraft to go overspeed, risking serious airframe damage. If you crash dive an airliner at cruising speed, very, very bad things happen. At cruising altitude safety envelopes are very small. The paragraph talking about direct law is, roughly speaking, what would have happened had this been a Boeing aircraft. (This isn't bad, it's just different philosophies of pilot responsibility. I'm not partisan on the difference like some pilots and am willing to acknowledge the benefits of both approaches; this one is an Airbus advantage.)
Nope, it wouldn't happen in a Boeing aircraft. This is a well-known Airbus defect that was a contributing factor to the crash of Air France Flight 447.
Boeing uses traditional controls. It's like a race car steering wheel on a big movable stick. It is really easy to see how the control is positioned.
Airbus uses a small joystick off to the side. It is force-based like a Thinkpad pointing stick, so it barely moves when you push on it. It isn't easy to see if the device is being pushed.
Here, a camera got jammed against the joystick. The pilot nearly made things worse by putting the aircraft in "direct law" mode because he couldn't tell that his own control was causing the problem. In the case of Air France Flight 447, the pilot's idiotic control input was not visible to the two other people in the cockpit.
Airbus will obviously brush this one under the carpet as well, because admitting that the side-stick control is hazardous would cause a need for drastic retrofit of all Airbus aircraft cockpits.
This is the partisanship I'm talking about, for what it's worth.
I've flown an A320 in a full-size simulator and my experience with its sidestick, even in that simulator environment, does not match what you are describing. The sidestick moves quite far and quite noticeably[0], and does not behave like a ThinkPad. The aircraft simply ignores drastic input, which might be what you're thinking of with that assessment?
Anyway, I'd imagine if I were PIC of this flight and my aircraft suddenly pitched down 18 degrees, the sidestick would be the first thing I'd look at in about the first 750 milliseconds. Not sure why that didn't happen here, but I tend not to judge pilots from accident reports, so I have't put a lot of thought into it. I can't imagine a scenario where any camera could conceal itself in that arrangement, unless the A330 has huge voids under the stick that the A320 does not. Even stopping short of judging the PIC, I have a hard time imagining an outcome where this is an Airbus control flaw, no matter the (as one can see, passionate) opinions on sidestick.
I don't think the sidestick is the problem, it's the lack of feedback and the fact that the system ignores conflicting input.
Boeing undoubtedly does that one aspect correctly: Regardless of the type of stick the two inputs are linked so if the captain is giving strange input it would be <i>immediately</i> obvious to the copilot. There is no reason Airbus couldn't implement the same system on their sidesticks.
The fact that conflicting input causes the system to ignore both seems like downright insanity and based on the assumption of one input being defective, rather than conflict between pilots. A simple fix might be to include a hand detection mechanism and ignore input from the stick that doesn't have a hand present unless it is the only input and is not overridden.
Bear in mind that the AF447 crash was primarily due to a junior officer not knowing basic flight principles. Visibility of the side stick to the other two crew (the captain had come back into the cockpit and was positioned between the first and second officer) is an issue, but I am not sure that central style yokes like Boeings have would have been a huge help to a crew already overloaded with an insurmountable amount of data to resolve.
There have been many cases of pilots turning the yoke the wrong way [1], or shutting down the wrong engine [2] which are levers in full view of both pilots and not getting immediately picked up.
In the case of AF447, the first officer (Robert) has to constantly ask the second officer (Bonin) what he is doing and why he is climbing, he's clearly worried about air speed (starting at 2:10:16 in [1]. Had the FO had direct control over the input, he could have felt that the SO was pulling back, and push back the column in neutral himself.
In the case of Crossair 498 (your first link), the copilot knew the captain was turning the wrong. Having linked controls, it is surprising he didn't take over. One of the conclusion was that there was a communication problem between the two pilots due to language barrier [2]. The copilot never gave any input to the controls, so the direct link controls were not an issue.
I'm not sure what how TransAsia 235 is linked to the issue discussed here, OP (tropo) didn't say that Boeing's approach prevents all pilot errors, but that mechanically linked controls have the advantage of letting one pilot know exactly what the other is doing, and give him a chance to intervene.
There is no question that computers doing most of the flying have improved reliability tremendously, but misinterpretation between computers and pilot of what the other is doing has been recognised has an issue and is something the aviation industry is focusing on right now.
> In the case of AF447, the first officer (Robert) has to constantly ask the second officer (Bonin) what he is doing and why he is climbing
The fact that this happens at all makes me think that there could have been a hardware malfunction or that something got wedged forcing the stick back ever so slightly. So FO says "why are you pulling back?" and SO doesn't even respond since his hand isn't on the control.
In this case the Boeing system wouldn't have stepped in, granted. But in this case the Boeing system also wouldn't have let things get so far so fast. It's much, much harder to accidentally jam a big control yoke so far forwards. It would also take more than the weight of a camera to do so.
The title should read "Airbus' automated flight control conflict resolution system saves plane from problem that Airbus' poor cockpit design allowed to happen in the first place" but that's not catchy.
I didn't look at the details, but keep in mind that AF447 lost the airspeed measurement. I'd guess the pilot is probably looking at the instruments, with wrong speed values, instead of the lever that never moves anyway.
I cited TransAir because it was a case where the throttle levers were in sight and easy access for both pilots, yet the co-pilot didn't pick that the captain was retarding the wrong throttle.
Not a criticism - I am a former commercial pilot, and I know just how hard identifying the bad engine in an asymmetric situation is - but it is an illustrative case that in a high workload situation, having visible controls that both pilots can see/access is not always as helpful as one would imagine.
The biggest advantage of the Boeing approach is that the controls are linked; if one moves, so does the other. Mismatched inputs are impossible.
I truly don't understand the Airbus approach... There is no method of resolving conflicts between the sticks that make sense... Sum them? Average them? Take the last seen input? Prefer the left side always?
I don't work in avionics, but in another field with two pilots. Most things are done by exclusive control meaning there is a selector for which station has the flight stick. But for shared control the way it works is the first station that has non-zero inputs takes exclusive control until they have zero inputs for a short period of time. Although in our case, the availability of two sticks is for hardware redundancy rather than pilot redundancy so the Boeing approach seems safer.
Of course there is a system, and it's sophisticated. Averages are taken in case of inputs on both, and there's a priority button to take precedence (so that the other is ignored). There's more to it than that, but of course there's a system and a lot of thought has gone into it.
In the AF 447 case, the left pilot did take priority, but the dude on the right took it back. Had the left one taken it and held it (by keeping the button pressed), the accident might have been avoided. So the system is hardly foolproof, and one can argue about the particular implementation of the interface and the merits vis-a-vis hardware-linked yokes, but any notion that Airbus haven't thought carefully about it is complete bogus.
I'm not saying it wasn't thought out, I'm saying that no amount of thought could result in a system that made sense in every situation... Having two independent control inputs means there is always room for subtle failure modes (either technical or operator related).
Multi-master is a tricky problem to solve. When it comes to flying a plane, sitting here in my armchair, I think the the simpler solution (linked yokes) is likely the better one.
Fair enough. I tend to agree, in particular as there was not only AF 447, but also the AirAsia from Surabaya to Singapore, Indonesia AirAsia Flight 8501, with basically the same failure mode.
So:
* KISS... but:
* Airbus engineers are certainly aware of it, though... but:
* they have thousands of Airbuses deployed out in the field though with the current tech, so it's unlikely to be changed anytime soon.
Is even more training the answer? Dunno.
My take:
Today's jets are incredibly well designed and safe. If something goes wrong, though, still better to have rather competent crew sitting at the pointy end.
Not knowing anything about it, it seemed to me from the article that the system takes mismatched input as an indication that the autonomous systems need to be engaged. Though if it subsequently decides not to engage, I'm not sure what you should do...
I could also imagine that a Boeing pilot generally has more respect of the machine (s)he's operating because of the direct control. The overall deadly accident rate in Boeing vs. Airbus, plotted over time, would be interesting.
Both aircraft are so incredibly safe with vanishingly low accident rates, that it'd be nearly impossible to suss out accidents due to the respective designs versus pilot error, weather, carrier procedures etc.
The silly thing is they could combine the fly-by-wire system with the boeing-style control system if they wanted to. There's no reason why not, other than the space in the cockpit I suppose. It's basically just a much bigger joystick, no?
All modern Boeings (>= B777) are FBW (fly by wire) and have flight envelope protection. A 787 will absolutely not go over-speed: it will automatically throttle back.
Also, Boeings have a traditional yoke. It seems like a lot of this trouble was caused by the fact a sidestick can easily be hit by something on the front of the chair.
Aren't the injuries trivial compared to the large risk of death? If seat belts are to be praised, it would be because they helped the pilot regain control, not for preventing some passenger bruising.
If any safety technology deserves credit here it's seatbelts. From the sound of it, nobody who was belted in was injured.