4 ms·

Should a lot of vehicles have fundamentally different controls now than when they were first invented? In particular, should we move away from controls which ma

by abeppu 6y ago

Should a lot of vehicles have fundamentally different controls now than when they were first invented? In particular, should we move away from controls which map onto how control surfaces or other physical parts move, and instead focus on controls that describe the desired motion? Quad-copters are pretty easy to fly, but they would be really really hard to fly if you were supposed to manage the thrust on each motor. I've never flown one, but I understand helicopters are challenging to fly, and controls do map onto actually actuating the swashplate and tail rotor. I don't understand how a gyroplane flies, or why pulling back on the stick causes immediate problems. But ... should autogyros have controls where you describe the motion you want, and let a computer figure out the actuation? I think there's an analogy to software here, where generally we're less error-prone and more productive if we can specify what we want by expressing higher-level, declarative relationships, rather than demanding that humans always think in terms of the low-level operations that will need to occur.

tjohns 6y ago
In aviation you want to avoid unnecessary complexity. You really want to avoid single points of failure. Most small aircraft are controlled via cables and pulleys because they're simple, can be redundant, easy to verify, and unlikely to fail. What's your backup plan when your flight control computer crashes? Or if you have an electrical failure? (There's an argument to have computers provide additional assistance on top of the manual systems, but we usually stop short of fundamentally remapping flight controls. And even then, we have to be careful otherwise we end up with situations like MCAS.)
- taneq 6y ago
  I think what you really want to do in aviation (or anything safety-critical) is reduce risk of failure. Avoiding risk of mechanical failure at the expense of controllability is simply pushing the blame onto the pilot. The general approach for critical hardware like flight controllers is triple redundant systems with two-out-of-three voting, driving redundant actuators. MCAS is a prime example of this NOT being followed, because it relied on a single sensor input to generate a significant control output.
williamscales 6y ago
From my limited reading, my understanding is that modern fighter jets work in this way. They use their aerodynamic surfaces in ways that are not considered "usual flight", i.e. when we see the very cool looking J-turn. But all the pilot is doing in that situation is manipulating the joystick, the aircraft flight computer figures out how to manipulate the control surfaces.
- mcguire 6y ago
  IIRC, aircraft with forward swept wings are very unstable and can only be controlled with fly by wire controls. The control surfaces are not necessarily doing what the pilot controls say they should be doing. https://en.wikipedia.org/wiki/Forward-swept_wing?wprov=sfla1 https://en.wikipedia.org/wiki/Forward-swept_wing?wprov=sfla1
- Toutouxc 6y ago
  It's called "relaxed stability". https://en.wikipedia.org/wiki/Relaxed_stability https://en.wikipedia.org/wiki/Relaxed_stability
jcrawfordor 6y ago
Airbus aircraft like the A320 are kind of like this. Roll and pitch is entirely fly-by-wire, and as I understand it, the basic scheme is that the sidestick position is taken as a desired rate of change and the avionics compute how to achieve that change using the control surfaces. This is somewhat necessary as a change in roll attitude is done using some or all of the slats, ailerons, and spoilers depending on the speed, pitch, etc. A cool consequence of this is that in the normal flight mode, pulling the sidestick full aft and keeping it there will result in maintaining best angle of climb. Under normal circumstances it is not possible to stall because the avionics will not allow control pressures that will leave the safe flight envelope. Changing the flaps, or more properly the configuration lever, not only adjusts flaps but also changes control parameters to modify the flight envelope for safer operation near the ground. The thing I found funny when reading into that is that, for all that effort for roll and pitch, the rudder is just hydraulic. I suppose there's just not that many ways you can fancy up yaw control.
- vvanders 6y ago
  There are rare cases where you want manual control, I remember in the A320 that ditched into the river Sullenburger said the limits prevent the best possible landing[1]: > However, Sullenberger said that these computer-imposed limits also prevented him from achieving the optimum landing flare for the ditching, which would have softened the impact. That said the cases for this are probably very, very rare. [1] https://en.m.wikipedia.org/wiki/US_Airways_Flight_1549 https://en.m.wikipedia.org/wiki/US_Airways_Flight_1549
- mcguire 6y ago
  When Airbus introduced its fly by wire systems, there were quite a few airframe losses caused by incorrectly handling the moded interface, including their test pilot dropping the A320 at the Paris Airshow. The crases were universally attributed to pilot error. If you have muscle memory trained in one system, you tend to revert to that behavior when under stress.

by abeppu 6y ago

tjohns 6y ago

In aviation you want to avoid unnecessary complexity. You really want to avoid single points of failure. Most small aircraft are controlled via cables and pulleys because they're simple, can be redundant, easy to verify, and unlikely to fail. What's your backup plan when your flight control computer crashes? Or if you have an electrical failure? (There's an argument to have computers provide additional assistance on top of the manual systems, but we usually stop short of fundamentally remapping flight controls. And even then, we have to be careful otherwise we end up with situations like MCAS.)

taneq 6y ago
I think what you really want to do in aviation (or anything safety-critical) is reduce risk of failure. Avoiding risk of mechanical failure at the expense of controllability is simply pushing the blame onto the pilot. The general approach for critical hardware like flight controllers is triple redundant systems with two-out-of-three voting, driving redundant actuators. MCAS is a prime example of this NOT being followed, because it relied on a single sensor input to generate a significant control output.

williamscales 6y ago

From my limited reading, my understanding is that modern fighter jets work in this way. They use their aerodynamic surfaces in ways that are not considered "usual flight", i.e. when we see the very cool looking J-turn. But all the pilot is doing in that situation is manipulating the joystick, the aircraft flight computer figures out how to manipulate the control surfaces.

mcguire 6y ago
IIRC, aircraft with forward swept wings are very unstable and can only be controlled with fly by wire controls. The control surfaces are not necessarily doing what the pilot controls say they should be doing. https://en.wikipedia.org/wiki/Forward-swept_wing?wprov=sfla1 https://en.wikipedia.org/wiki/Forward-swept_wing?wprov=sfla1
Toutouxc 6y ago
It's called "relaxed stability". https://en.wikipedia.org/wiki/Relaxed_stability https://en.wikipedia.org/wiki/Relaxed_stability

jcrawfordor 6y ago

Airbus aircraft like the A320 are kind of like this. Roll and pitch is entirely fly-by-wire, and as I understand it, the basic scheme is that the sidestick position is taken as a desired rate of change and the avionics compute how to achieve that change using the control surfaces. This is somewhat necessary as a change in roll attitude is done using some or all of the slats, ailerons, and spoilers depending on the speed, pitch, etc. A cool consequence of this is that in the normal flight mode, pulling the sidestick full aft and keeping it there will result in maintaining best angle of climb. Under normal circumstances it is not possible to stall because the avionics will not allow control pressures that will leave the safe flight envelope. Changing the flaps, or more properly the configuration lever, not only adjusts flaps but also changes control parameters to modify the flight envelope for safer operation near the ground. The thing I found funny when reading into that is that, for all that effort for roll and pitch, the rudder is just hydraulic. I suppose there's just not that many ways you can fancy up yaw control.

vvanders 6y ago
There are rare cases where you want manual control, I remember in the A320 that ditched into the river Sullenburger said the limits prevent the best possible landing[1]: > However, Sullenberger said that these computer-imposed limits also prevented him from achieving the optimum landing flare for the ditching, which would have softened the impact. That said the cases for this are probably very, very rare. [1] https://en.m.wikipedia.org/wiki/US_Airways_Flight_1549 https://en.m.wikipedia.org/wiki/US_Airways_Flight_1549
mcguire 6y ago
When Airbus introduced its fly by wire systems, there were quite a few airframe losses caused by incorrectly handling the moded interface, including their test pilot dropping the A320 at the Paris Airshow. The crases were universally attributed to pilot error. If you have muscle memory trained in one system, you tend to revert to that behavior when under stress.