...robots obey their limitations...
Well, we are ALL constrained by our limitations - by definition - are we not? !
No, we are not constrained, we operate beyond our limitations on occasions and rely on luck whether the outcome is as we wish. That is because of what you could call "intelligence", we do not understand absolutes ie nothing is absolute in our minds. If something is stated as an absolute, there will be a human somewhere that says "what if .....".
I think you mean robots will reliably follow instructions. Generally they do, until a cosmic ray flips a bit somewhere and it goes crazy! Once a program jumps to the wrong area of memory it can start interpreting data as machine code, and then god knows what it will do!
We rely on millions of things not to break all the time - can anyone predict when a brake seal will fail on a car? (noting that most brake seals on cars are not protected by dual circuit brakes).
Road vehicles are possibly the only form of powered transport that have have not been used autonomously for many years. Trains, ships, aeroplanes all have autonomous systems in use.
The cost of adding independent safety systems to check the autonomous system is working correctly is minimal, I doubt if it would add £50 to the cost of the car.