[stuartloxton]

Amusing - but a car with 2 passengers going at a conservative 30MPH is very different to a plane flying 9KM over the ground at 567MPH.

I'm just hoping that in 80 years I won't be alive to see my comment becoming horribly wrong, with people laughing at it while a processor flies them across the world.

[apl]

Amusing - but a car with 2 passengers going at a conservative 30MPH is very different to a plane flying 9KM over the ground at 567MPH.

You're right: building an automatic car is much harder. It's counterintuitive, but speed or altitude are irrelevant if the problem space is sufficiently simple. Additionally, risk doesn't scale linearly. If a plane crashes at 800km/h, all people on board will die; if a car crashes at 130km/h, you'll see a very similar result.

[stuartloxton]

That's true for the occupants of the vehicle but the dangers of those outside are greatly different. I'm not saying that planes will never be flown autonomously however there are a lot of hurdles including convincing the public.

If a car when parking is out by 1% it's maybe sticking out of the space and causes other people to have to park badly too, if a plane rounds incorrectly or a sensor plays up and is 1degree out (so even less than 1%) it lands onto of a terminal filled with tourists.

Even as a professional programmer I tend to trust unknown programmers less and less, bugs get uncovered too late, shortcuts taken... online e-commerce fine, cars and busses - maybe but scaling up the trust and risk involved is difficult.

[Figs]

Planes were capable of being flown autonomously in the 1980s. In particular, the Soviet's space shuttle, Buran, performed unmanned orbital maneuvers and landing during its 1988 test flight.

http://en.wikipedia.org/wiki/Buran_%28spacecraft%29

So, the technology to do these kinds of things has already been done before (though unfortunately neglected!). I think the main challenge is just convincing people that it's safe enough that they will be willing to buy tickets.

[eru]

I guess the biggest problems will be the legal framework and the power of unions. Technically flying a plane should be easier for a computer than driving a car.

[WalterBright]

Where a computer fails is when judgment is required. For example, you have an in-air emergency, and are going to have to crash land. What do you pick to crash land on to?

Consider Captain Sully's recent decisions in just such an emergency.

[winthrowe]

Being a private pilot with a lapsed license, I can't possibly see that emergency landing area selection is a problem worse than navigating a street vehicle in an urban environment. Taking all the GIS databases around, I can't imagine emergency landing site selection would be a huge roadblock.

[eru]

And for lots of emergencies you can probably do remote flying from some call centre in India.