Hacker News new | past | comments | ask | show | jobs | submit login

> E.g. London-Amsterdam could be dozens/hundreds of 6-12 seater electrical planes

As long as they don't all need 2 pilots. I'd really like to see self-driving planes, must be easier than self-driving cars surely? (And, yes, I would fly in one)




Flying a plane is easy. Autopilots have been a thing for a long time. Landing a plane is a bit tricky but doable. The value of human pilots though is in the case of malfunctions. That's the main problem. A failing safe-driving car has always the option of stopping on the side, calling home and lighting the warning lights.

A plane has to find a way of landing safely first. I am not versed enough in aeronautics to even begin to understand how hard it is to reach human-level in mayday situations.


Under normal circumstances a plane can totally land itself at an airport equipped with ILS IIIC. If you’re landing in a corn field because your engine stopped spinning (generally a bad day) the situation is a bit different. This is why I think fully autonomous planes carrying humans won’t happen for a really long time. The stakes when shit goes sideways is really high and it takes skill and a bit of luck to recover from that. Computers don’t yet have that.


Many small business jets are single pilot operations. I think the FAA has 6 passengers as the upper limit for this currently (but I could be wrong). Of course rules can be changed when technology makes this possible. The main limitations for self flying planes are non technical and related to regulations and procedures designed for human operated planes. IMHO, much of that could be solved with today's technology even but it would require disruptive changes that don't quite make sense yet. Once technology catches up it will make more sense and these things will start changing.


>As long as they don't all need 2 pilots. I'd really like to see self-driving planes, must be easier than self-driving cars surely? (And, yes, I would fly in one)

The issue I think is that airplane autopilots rely significantly on sensors which occasionally fail. That means self-flying planes must be able to cope with sensor failures which generally means falling back to visual cues. Self-driving cars that rely just on cameras (rather than LIDAR) are a point of debate regarding their feasibility. Planes that would have to rely on just cameras likely have the same issues.


There is intermediate possibility, the pilot does not have to be on the plane. We already fly drones remotely.


>I'd really like to see self-driving planes, must be easier than self-driving cars surely? (And, yes, I would fly in one)

I wouldn't, if it was made by Boeing. Just look at how badly they fucked up something as simple as their MCAS system, making it only use input from a single sensor. That kind of incompetence isn't going to yield a safe plane that flies by itself.




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: