Tourist,
Good post. The issue of drone crash rates is a real one, and will need to be addressed. Operating over essentially empty deserts in completely unopposed airspace has allowed the design tradeoffs to focus on endurance at the expense of reliability (so no back up generators, reciprocating engines, etc).
The fact that the US has (by comparison with any other nation) an essentially unlimited defence budget has also allowed them to tolerate the expense of these loss rates.
That will need to change as drones move out to other nations, and need to operate in more congested or challenging areas. While pilots aren't at risk, those on the ground most certainly are.
I would offer the opinion that designing more usual levels of redundancy into drones won't be straightforward, especially at the smaller end of the market. They are built more like powered gliders, with small weight margins, limited electrical capacity and very limited internal space. They have little margin for inserting additional systems. Further up the line, it's still not easy. The Global Hawk programme suffered simply epic increases in cost and delays as the original design concept had to be recast to reflect the very high cost and sensitive payloads being carried. (And there were still a number of losses).
To be clear, I'm not 'anti-drone'. They have, and will continue to have, huge military utility. But I do feel (it's just my opinion) that the initial optimism about drones taking over the world will be tempered by the real world difficulties of building reliable and effective flying machines - that's a challenge whether the vehicle manned or unmanned.
Best regards as ever to those getting the beasts in the air and keeping them there,
Engines