Oh, I'm so going to get flamed for this....
Hello there - I'm a geeky UAV engineer. The UAV bit is recent, the other bits - less so. I honestly believe that in 2100 most of the aircraft that don't need people in them, won't have people in them. Everyone who says that CAA doesn't allow UAVs in unsegregated airspace because we don't know how to make Sense & Avoid work is ENTIRELY CORRECT. However, geeky engineers & scientists have a habit of making stuff work, in the end, that other people didn't. It's why we're geeks.
On the other hand, we're also good at making things like major disasters & global warming, but you like your PC & light bulbs don't you? And that aircraft you're so fond of flying? We did that.
I'll give a couple of definitions, so you know what I'm talking about
:
1) Every UAV in service today is a Remotely Piloted Aircraft. If the comms link goes down, it might try something sensible, but fundamentally at best it'll have a 'run home to mummy' function & require that everybody else gets out of its way. I'm not talking about these as they are a near term solution & will be gone by 2020 or sooner. Example: The GA Sky Warrior can be flown from an Apache & has autoland where Predator didn't - so the automation level is going up all the time, even in the same airframe.
2) Autonomous UAVs will not be NOT pilotless*. They will do lots for themselves, but they will have an operator who will understand airspace operations - & that the UAV can go to for permissions & help when required.
* - OK, so the ones that will be smaller than a pigeon** or insect sized will be. But you're happy to have dogs help in IED sniffing, so we figure you'll cope with small vehicles with dog/horse levels of intelligence by 2030. Ish.
** - because any bigger & they're a birdstrike hazard & will need airspace understanding!
The traditional anti-UAV arguments at this point are something along the lines of : "Well, you can't programme in all circumstances & a good pilot will save the aircraft" and ye olde chestnut:
"How many times has your phone/PC crashed this week".
The chestnut first: Be honest, how many times has it? Probably less than a Windows 95 PC did. But more importantly: How many times has the safety critical software running on safety critical hardware in A320/330/340/380, B777 crashed this week? How many times has the Typhoon/Gripen/F-22/Tornado/F-16/F-18 fleet been grounded this week because of system crashes? Yes, there will be a few gripes in the initial phases - but the experiences of the 90s have made flight controls guys very sensitive on the matter. Please acknowledge that Microsoft do not, yet, write safety critical software. When they do, they'll be subject to the same rigour & legal ramifications as the rest of us that claim the term.
As an aside - this is why GA do not have and have not sought civil certification for Predator or Reaper - you have to design in sufficient resilience for civil certification from the beginning & have legally supported evidence that you have done so. Or re-design everything at great expense.
Now then: "what about the one in a million..." Honest answer... In a manned aircraft, I care about the one in a million, because I have to save the man. In an unmanned aircraft I've only got to worry about the guy in the other aircraft that is about to hit me or the people on the ground I'd squish.
Remember what I said at the top: no person if one isn't necessary. So an unmanned transport aircraft or JDAM dropper - go UAV. Unmanned recce platform with 5 day endurance (unrefuelled): go for it.
Unmanned passenger aircraft - why? You've spent the power & weight on life support & internal volume - why not add one or two more slots for aircrew.
Consequence: Even if I couldn't get to the 'one in a million', I'd cope because it's actually 'one in 10 million' when I don't have a life to save all the time. The aircraft probably costs 50% what the manned equivalent costs & it probably flies 3x as many hours as a manned aircraft, so the economies work. And without corporate manslaughter or lack of duty of care provision, I may even have adequate insurance cover & only have to pay the excess... Just sayin'...
And in terms of people-on-the-ground avoidance, that's a different problem from sense & avoid - plenty of useless dirt & empty sea in the world to start off with. The BA B777 & Hudson A320 are actually good examples of where an automatic system, that did NOT have to keep pilots in the loop would have coped quite happily. Both are control & trajectory management problems that computers are really good at.
If you a put a UAV in the Hudson example, it's likely that we wouldn't see the point of it gently landing in the water - since the aircraft would not be flyable again anyway & I have no occupant to keep comfy. So we'd try to make it spear into the water to contain the damage. I know people working on exactly this problem for UAVs.
Anyway, happy to be queried - within the bounds of "won't get me fired". Some 'love the geek' would be nice. But, I think, unlikely...
