This topic comes up every now and then amongst pilots. Aeroplanes are getting more and more advanced and the technical requirement for human pilots being in control is diminishing. Let’s say that sometime in the near future, the infrastructure and technology was readily available to be able to remove the pilot from the aeroplane. It is shown that accident rates would be lower. There would be no pilot error accidents but there would still be occasions where the aeroplane couldn’t come up with an out-of-the-box solution to a problem. In this future world, a Sullenburger accident would have resulted in fatalities but Air France passengers would have made it to their destination never suspecting something was wrong (the autopilot would’ve set known attitude and thrust values on detection of the airdata fault).
The question is, would you step on to an aeroplane knowing it was just George the autopilot driving up the front?
Key points.
No pilot.
Accident rates are lower with pilot errors removed but pilot-did-good incidents also removed
The technology is such that there are no new failure modes, it’s all the same problems we have today but they are handled in a preprogrammed way.
Absolutely. But not until the technology was proven in service.
Incidentally there is no reason in principle why decent enough AI wouldn’t have landed in the Hudson. In fact that would have been the expected result - AI would immediately have concluded it couldn’t make an airport and so would have looked for the best thing to hit. If the AI of this hypothetical aircraft couldn’t even do that then it ain’t good enough for me to fly on.
I don’t want it to be too futuristic. To be able to do what Sully did it’d have to have the coordinates of the river programmed in and it would have to be able to “see” that the river was clear with no logs, people, boats etc in the way. That capability is beyond what I’m thinking of. My pilotless aeroplanes can takeoff and land from any suitable airfield with an ILS. It can automatically shut down and secure engines when necessary. It can isolate faulty systems and automatically switch to backups. It can respond to a depressurization and conduct an emergency descent. In essence it can respond correctly to the failures that current aircraft have checklists for but it can’t cope with unforeseen failure combinations.
Edit: I think the problem may be that nobody would fly in it until it had been proven in service and therefore it would never get a chance to be proven.
What terentii said. As a side, I have travelled with the Dockland Light Railway in London, both with and without a driver, and I must say that it’s a much smoother ride with a human at the controls.
lol then you don’t understand how the DLR works. The only way it can be completely manual is by being limited to a very slow speed and this is never normally done.
The vast majority of times that there is a “human at the controls” he does little more than close/open the doors.
Sorry, but I know very well how the DLR works and the times I have travelled with it when they have driven manually it has been with normal speed. The only time when it was very slow, because they were limited to just one track, it was on automatic.
Eh. There have been a lot of crashes caused by misprogrammed autopilots. (Sometimes the meat pilot walks away, sometimes not.)
And sometimes ILS breaks down. What then? What if your engine falls off and you need to land at a small airport with no ILS? Or on a grass levee?
I think at the present time, based solely on my layman’s opinions and having watched a marathon of Mayday episodes on YouTube, is that the advantages of humans in the cockpit still outweighs the risks. Maybe when autopilots get vastly better, that will change.
Yep there will be a certain amount of crashes due to equipment failure but there will be no pilot error crashes. One of the terms of the hypothetical is that this balances out in favour of not having pilots.
It doesn’t matter how it taxis to the runway. It can have guides on the taxiway centerline or something.
The question boils down to: Are you prepared to fly in an “auto-plane” if it is demonstrably safer than a piloted aeroplane but not 100% safe?
In that case: sure. I’m a big fan of automation. Automated processes have regularly proven themselves to be safer and more reliable than humans. In the world of transportation, there have been several successful implementations of fully automatic trains, some of which I ride regularly.
But I think we are many decades away from achieving the technology to do the same thing with airplanes.
I guess the big thing for me is that for most of the failure modes for this “auto plane” you are going to have possibly HOURS of hopelessness, knowing that the computer can’t save you - and possibly you’ll be in circumstances where any moron who doesn’t necessarily need to be a trained pilot could save the day (for example given you are not giving the aircraft any AI, just losing its NAVAID database would be enough to render it incapable of landing). Pilot error is usually CFIT of one form or another, a much more pleasant way to go psychologically speaking.
A more interesting question - how about if you had these pilotless aircraft but you also had say one pilot for every ten planes who could remotely intervene in things? I would probably be fine with flying on those, even with today’s tech, so long as he could (for example) land just from the nose camera beamed back to base.
That’s not that difficult and I wouldn’t be surprised if that hasn’t already been started. I don’t believe the FAA is doing anything like this for their charts just yet.
Unless it’s a CATIII ILS there is still plenty of margin for error. You still need to visibly sight the runway to land. They are developing RNAV and RNP charts that are supposed to be much more accurate.
Right now I wouldn’t fly in one, too many little things that we still need a human for. There are lots of little things that go wrong that pilots can manage to take care of. How often do we see landing gear problems and such?
There’s nothing special about a CATIII ILS, an aircraft with auto land can technically auto land on a CATI ILS, its just not legal. Part of the infrastructure would be certifying more CATIII approaches.
As I said earlier, whatever failures it can’t handle are made up for by pilot error crashes that don’t occur such that the new system is safer, that is a condition of the hypothetical. It doesn’t matter what the technicalities are, you are statistically less likely to die on the autoplane, do you fly on it?
The logical answer is yes, but I’m interested to know whether there is something comforting about having a real person in control so that you’d rather be less safe at the hands of a person than slightly safer at the hands of a machine.
I don’t think I would fly on it but I’m not sure I can say why exactly.
I think the real question would-be flyers are going to ask themselves is: Do I really really believe that this robo-plane is safer? No matter what Professor Pearse or whoever says?
ETA: And I have a hunch that a lot of people will fear just what Confused dart cum suggests, above: That robo-planes may have some very nightmarish failure modes, whereas people imagine that pilot errors will more likely be just one fast splat and it’s done.
And BTW, wasn’t that Air France splash-down cause by some serious electronics/controls failure too?