How does autonomous aircraft fly through thunderstorms?
Current weather radar pictures are synthetic in most modern airliners. This means that the computer has already looked at the raw data, analysed the information and presented what it believes to be useful info about it to the human pilot.
Now, we may well have opinions about how well it does this, but either way, the computer is already in the loop.
That is correct it generates a composite synthetic of what it thinks is ahead. However this is not always accurate. We can look out the window and experience will tell us what is dangerous and what isn't. Or alternatively a small light green return with a bolt of lightening coming out of it is probably best avoided even though radar says its fine. How does a autonomous aircraft know what it doesn't know?
How do autonomous aircraft land in gusty 40knot crosswinds?
How do autonomous aircraft land with shifting winds?
As previously discussed and agreed by John Farley, the answer is better than humans since the Comet.
Well from my experience this isn't true. I can always beat the autothrottle/autopilot in these type of conditions as it cannot anticipate. Similarly why is the manual landing crosswind so much higher than the autoland one? And will this be the same in autno mous land.
Shifting winds on final is something else that current aircraft don't do well. Especially tailwind to landing in headwind scenario.
How do autonomous aircraft handle data input failures or data corruption?
(ie airspeed indications are no longer reliable or false sensory inputs ie stall warning goes off incorrectly)
Modern airliners are already totally dependent on software to fly. The computers are between the pilot and the controls whether they are autonomous or not. If data being corrupted is a problem, then the problem is already with us.
Added to this, it has unfortunately been shown that human pilots don't necessarily deal with such a scenario anyway.
Once again not true. I know off the top of head 5 data corruption failures where humans hand flew aircraft with all sorts of incorrect data being presented realised it was false data and ignored all the computers and landed the aircraft safely, non eventfully. Blocked Pitots, False Stall Warnings, False information, Total computer shutdown, etc etc.
So how will an autonomous aircraft figure out what is erroneous input and what is real? Computers are ultimately only able to compute what infomation they are given. No matter how advanced the old GIGO analogy still applies.
I guess similarly how will a computer know when another goes rogue? My assumption here is that two independant computers will be running the show but how will they know what is real and what isn't?
This doesn't even consider the extreme failures over the past 30 yeas where humans have really saved the day.
I would imagine most of the above problems could be solved however the data corruption issue could be a bit of a achilles heal.