But you should also consider where it is OK for delivery trucks to crash, and whether a drone crash or truck crash is more likely to cause damage or injury.
Hence overhead non-highway surface streets. Drones are typically fairly small, smaller than delivery trucks. And hopefully the drone operators can figure out how to avoid power lines, overhead trains, etc. Wind could blow a non-functioning (or even worse, incorrectly vectored) drone off the street as it falls towards the ground. A market test could start off by declaring streets OK for overhead drone activity.
I bet if the FAA opened it up to cities to decide whether it's OK for a drone to operate commercially above their streets, you'd get dozens or hundreds of cities move to allow it. Starting with Seattle.
I'm not sure why you think it would be okay to fly over non-highway streets, where presumably there will be pedestrians. Already there has been an injury caused by a drone falling on top of someone (http://www.bbc.com/news/technology-26921504).
Regarding avoiding power lines: is that actually feasible? Either you have to program a GPS route from the warehouse to the delivery address avoiding all obstructions, or you have to use some kind of radar so that you can detect obstructions even in fog (or maybe just not operate in low visibiliity). It's not a trivial problem, and I'm not sure if the proponents have really thought it out.
PS, if you're going to downvote my comment, at least explain why.
