Originally Posted by
riff_raff
If you think about it, even a fully autonomous aircraft still relies on a control system that functions based on a set of instructions created by a group of humans that designed the controls using their own best judgement on how to deal with any particular situation they thought the aircraft might encounter.
This is not strictly true with a neural net, and even in the areas where that comment may be valid, remember that the humans making that plan have the benefit of time and a testing process. That is exactly why we have ECAM and QRA in the first place. They may not be perfect but they are a darn sight better than working on the fly.
Originally Posted by
riff_raff
This approach works very well in most cases, but there are the rare situations where a skilled pilot with the ability to make split-second decisions can do a better job.
Human are not good at split second decisions.
That is what computers are good at.
Humans are good at events that are entirely new and unforeseen. Black Swan events.
That is what computers are currently bad at.
How many accidents do you think are Black Swan and how many are the same old accident again and again....
Sully's adventure is usually quoted about now.
His example is actually exactly what an autonomous aircraft would be good at.
An aircraft will obviously know exactly where it is (GPS/INS) and it's energy level. Those, combined with a simple performance data set will give it it's glide range.
It now knows if it can make the nearest airfield.
Not guesses, not wonders.
It knows.
If it can, it goes to the field. (I have seen rumours that it may have been able to. I don't know)
If not, then it makes the same decisions that a human would.
Is there a river/lake in the database?
It will do that near instantaneously whilst putting out a mayday informing exactly where it is going to land, completing the ditching drills, briefing the cabin crew, telling the passengers to brace etc etc.
None of that is technologically too challenging.
There even exist flying today on an optionally manned Blackhawk a system which if the river was not there will scan the terrain for the least worst forced landing area.
Originally Posted by
riff_raff
When it comes to ensuring the safety of commercial airline passengers, the added cost of having two qualified pilots in the cockpit is money well spent.
Which is it?
Is an autonomous system to expensive to produce as some would have it, or an attempt to save money by binning pilots?
I think it is not unreasonable to make this statement.
Every automation brought into airline cockpits so far have led to the reduction in reliance on pilots, and an improvement in safety.
What makes you think that further steps towards the logical conclusion will be any different?