Actually autonomous ones are as far as I know only good for static targets (effectively they are reusable cruise missiles).Close air support or deciding whether to engage an enemy airplane or not gets tricky.
There are different types of UAVs "drones" for a more buzzword-y term, for different uses. The bigger boys such as the Predator, Global Hawk, etc use satellite navigation. There is most definitely a "lag" of sorts for those types of UAVs. The Shadow and the Hunter however, are line of sight. It limits their range severely (I got one out to 166km from the control station), but they are just as dangerous, albeit smaller and with less endurance for extended missions. It doesn't matter, they are cheap and you can launch lots of them :)
I've got 480 combat flight hours from Operation Iraqi Freedom II, from 2003-2004. We CONSTANTLY were targeting moving things and never had any real issues with it. You just use the right drone for the mission at hand. The military has dozens of different types of drones, but only the bigger ones make the news on a regular basis. The Shadow is being phased out for those types of missions in favor of the MQ-1C Gray Eagle.
Now for air/air stuff, they call those UCAS (lookup the X45-B / X47-B), the tech is still fledgling enough to need another 5-10 years before it will ever become a reality. On that, there simply isn't good enough software... yet.
UAVs are great against opponents with limited electronic warfare capabilities, but wouldn't they lack credibility against China or Russia? They'd simply jam them and that'd be that, yes?
Blocking the entire RF spectrum all at the same time is possible, but unbelievably difficult. When you're hopping a random # of MHZ every few dozen microseconds, it is super tricky.
I guess that was what was behind the story of insurgents grabbing drone video feeds.
Insurgents Hack U.S. Drones http://online.wsj.com/news/articles/SB126102247889095011
As I understand it, that's so you can show them to anyone quickly; if encrypted, various security issues are invoked.
It was considered to be an acceptable trade off because the greater part of their usefulness is so ephemeral. On the other hand, an adversary who routinely monitors them could figure out a lot of things about our general methods.
Manned planes are the cavalry of the 21st century. They're glamorous and dashing, but also insanely high maintenance.
On-board AI will only get better. And you only need one kill from a disposable unit to take out a manually-flown enemy.
The fact that this terrifies so few people terrifies me.
Pretty soon, some one will ask "how many teachers could we hire" or "how many roads could we fix" for the price of "the war" ...
The longer range ones really can use it to increase hit probability. Or so I've read is true for the Phoenix and AIM-120 missiles, classed as "fire and forget". and the newest Standard missile uses a AIM-120 derived head with a bigger antenna. And I'm sure still depends on mid-course corrections (for that matter, I've read the current models self-destruct if they doesn't get guidance fairly soon after launch).
The AIM-120 uses a datalink to guide the missile close to the target, after which the missile's internal radar emitter goes active, and the missile locks onto the nearest target that it finds. This is because the radar on the missile is not nearly as powerful as the radar on the aircraft. So it's not truly a "fire and forget" missile, at least if a high probability of kill is desired. I believe the Phoenix is similar.