If you compare this to the problem of driving a car in an urban environment it starts to look really simple by comparison. Put the right sensors on the UAV and a little intelligence (avoid hitting anything except the ground when landing - and that not too hard). Note that vision is not necessarily the best choice. Lidar gives better ranging information and can scan across all quadrants.
Having spent some time in the pilot's seat of a small aircraft, I well understand the challenges involved in keeping away of any and all craft sharing the near airspace. It's not easy for an alert human. Today it's not easy for a robotic craft; maybe not possible.
However, that will change soon enough. If a cheap digital camera can pick out a face and wait to release the shutter until that face is smiling, the ability to see and avoid a moving object in the air is just a matter of applying the technology in that direction.
Relative to the cost of the drone, 360 x 360 degree video coverage is not that expensive. Add in a number of ARM Cortex processors to do the recognition and you've upped the cost and the weight, but not substantially.
Ultimately, even if I'm the one in the pilot seat, I would prefer that my and the other air vehicles around have a more reliable vision system than does a human. Machines fail, but so do people.
It isn't even an issue of fear, or lack thereof. A drone can pull maneuvers that a human can't survive, and it has the "reflexes" to take full advantage of those maneuvers. The drones in the video recalc and adjust control surfaces 600 times per second with (literally) inhuman precision. How does a human fight that?
Yes, imagine if a jet fighter with a pilot on board get involved in a dog fight with a drone, the drone has no fear of death and the pilot does, this is total asymmetric warfare and the human pilot will inevitably lose.
Current military drones can fly autonomously and so accurately that a human pilot simply cannot compete. Check out http://arstechnica.com/science/news/2012/03/robots-swarm-the-stage-at-ted.ars for the state of the art on the research side on these guys. If I were a fighter pilot I would be planning my retirement right about now before I had to face one of these in the sky.
Not to mention the requirement of absolute control over the RF link which will be next to impossible from and RF engineers perspective and the technology and control that is available today.
This will open up a new arena for hackers.
Can you imagine the RF equivalent to the internet group 'annonymous' where they break in to the RF link and start messing with these drones ?
Also it seems they aren't that advanced that they are 100% safe from crashing randomly.
They will need to be 1005 safe from incidents like this otherwise the lawyers will be dictating how they will and will not be used.
See the following
I live near three major airports and a hospital. The amount of daily air traffic is astounding. (I've even seen the shuttle and International Space Station fly over.) Add to the mix the medivac helicopter making daily runs, the county police copters mostly wasting aviation fuel and the traffic plane telling us what we already know (all roads are backed up during rush hour). It's very hard to see how you fit drones into the mix unless they are flying at tree-top level. The best use for "surveillance" drones seems to be in remote areas, away from populated areas and airports.
Absolutely understandable in a battlefield situation (cost / benefit makes sense) where our soldiers' lives are saved (as are innocent civilians on the other side). My concerns relate to having a half ton drone loitering over our school playground reporting on traffic, tracking shopping mall shopping patterns, or monitoring whose lawns need mowing for the lawn service. I'm presuming these are the kind of things for which the FAA is exploring regulating "eye in the sky" domestic applications.