Autonomous aircraft

http://www.telegraph.co.uk/business/2017/08/07/computer-speaking-now-cruising-580mph-altitude-36000ft/

 

“This is your computer speaking. We’re currently cruising at 580mph at an altitude of 36,000ft….”

 

It sounds incredible but airliners without a human at the controls could be flying passengers through the skies with in a decade - saving airlines billions by doing away with pilots and cutting ticket prices for passengers.

 

Research by analysts at UBS claims the pilotless aircraft could generate $35bn a year in savings for airlines.

 

 

 

And I just love this part:

 

but by making aircraft safer by having them controlled by computers which are less likely to make errors.

 

 

BAHAHAHAHA!!! ;)

The problem with autonomy is that it copes with the expected very well indeed, but not with the unexpected. Autopilots have been with us almost as long as flying itself in one form or another, but always with an 'off' switch, and for good reason. Whilst it is very true that sometimes human guidance can be flawed, it can also respond intuitively and with great foresight. After all, remember that incident on the Hudson River? An airliner captain with long experience and finely honed flying skills brought his stricken airliner onto the river safely enough for his passengers to get out unharmed. It's hard to see AI responding like that however sophisticated it may be, and even today, autopilots are not 100% reliable. With aircraft sizes inflating toward carrying the population of a small town (The Antonov 225, whilst a unique aeroplane, is capable of carrying a passenger load of 1500 people if it was so equipped) and the reliance on autonomy, the magic of flight seems a good deal less inviting to me.

***but not with the unexpected***

 

Really is the END of the story of why autonomous aircraft will NOT become the norm.

 

The Hudson river malady is a perfect example.

 

I would not get on one.

Yeah Mr. Autonomous! Lets just see how you handle a couple of bird strikes right in both engines!

I think that it may happen some day.

 

First it will happen with cargo flights: No passengers to kill.

When they establish a good record people will want to do the same for passenger flights more and more.

Successes with autonomous cars may also help to make autonomous flights more acceptable.

 

Then it will become a matter of statistics to decide if passenger carrying flights can be done by computers as well.

What has more fatalities at the bottom line? Human or computer operated flights?

 

And how often we have "miracle on the Hudson" type incidents?

 

While a computer may not be able to truly invent something on the spot, it sure can think fast.

It could carry out the planned flight and simultaneously compute what to do if , what to do if that and switch as soon as something occurs.

I've yet to own a computer I trust to save a text file so have great reservations about autonomous cars, much less autonomous aircraft.

 

Well, AI improvements (in the real world) are are advancing exponentially, so autonomous vehicles and air transport are definitely predictable for the future.

But, they will only be accepted by the public and corporations only after a well-documented and proven track record of low incidents and accidents...

 

Right now though, I would not dare step into an Uber, et al without a living and breathing driver.

I've yet to own a computer I trust to save a text file so have great reservations about autonomous cars, much less autonomous aircraft.

 

[ATTACH=CONFIG]199112[/ATTACH]

 

Keep in mind that software developed for aircraft is very different than your desktop computer. The hardware is fixed and you don't have third parties writing poor quality drivers or applications that can crash the system or otherwise cause problems.

The Hudson river malady is a perfect example.

 

That would actually be a pretty straight forward case for an autonomous aircraft. Combining an internal database of terrain and cities, much like Google Earth has, with integrated optical, IR and radar data similar to what is being developed for the F-35 would allow an AI to pick a suitable landing spot and bring the aircraft down. Could probably react and process everything faster than a human too.

 

A more tricky example would be Air France 447 where the problems started because the airspeed readings were inconsistent. On the other hand, it was the pilots that ultimately stalled the airliner.

 

The Asiana crash in San Francisco was caused by the pilot, and so was the Air Canada crash in Halifax a couple of years ago. And Air Canada just about had a really bad accident recently in SF caused by the pilots confusing the taxiway for the runway. Relying on humans isn't any guarantee either.

 

Qantas flight 32 is probably the best example if one wants to look at the need for creative troubleshooting.

 

AI development is really only just ramping up and bears little resemblance to what current autopilots do. The US Navy has had a UCAV land itself on an aircraft carrier, for example. 5 to 10 years may be a little soon, but autonomous aircraft will be possible sooner than many expect.

***autonomous aircraft will be possible sooner than many expect***

 

In yer fantasy filled dreams.

 

People will NOT set foot on the things.

Maybe they will, and maybe they won't. And it could take a while to happen. It wouldn't be the first time people resisted change that worked out for the better in the end.

 

http://www.npr.org/2015/07/31/427990392/remembering-when-driverless-elevators-drew-skepticism

 

Let's not forget about the pilots that fell asleep or were distracted and flew right past their destination. Or the ones that landed at the wrong airport. And worse, the one that deliberately flew into a mountain side. Humans are far from perfect.

 

As for fantasy filled dreams, we have the technology right now. It will probably take people getting used to autonomous cars first, but I don't think it's nearly as black and white as you seem to think.