Technology tends towards mechanization. The history of warfare may as well be synonymous with the history of technology (certain aspects, anyway), so it's fair to say that as some point, wars will be fought between operators hundreds of miles away from the battlefield that the war is taking place. Here's the deal, though.
The robots won't be androids. (I mean, technically some people would stop calling them robots at that point, but clearly not you if you're using BattleBots as an example of a robot.)
There are numerous problems with androids, and even if you manage to solve them all, you get the real kicker of the problem - androids just aren't efficient. Humans are great multitaskers, a natural jack-of-all-trades. But we aren't specialized to hunt and kill. So these bots aren't going to be android soldier a la Terminators. They also aren't going to be giant humanoids either (square-cube law). Instead, they'll be tanks, drones, basically all the vehicles we have today, just have their insides replaced with electronic operators rather than flesh and blood. That's what you'd be looking at. And they'd have guns and missiles, and more guns and more missiles. Because in war, any problem which you can't solve with overwhelming firepower just means you haven't used enough overwhelming firepower.
The problem (amongst many) is that you're losing specification. Humans can be instructed to accomplish objectives like hostage rescue, taking strategic targets with minimal engagement, and rallying indigenous populations. Robots can't do that. We'll probably never get a robot with the jack-of-all-trades of a human, but we'll be able to replace large-scale conflicts with robot armies.
Counter-attacking would be interesting. EMPs might seem tempting, but drones can be designed with shielding. Jamming and hacking might work, but if your opponents have better hackers, and program the 'bots to turn into merciless killing drones on innate programming when the command frequency is lost that could backfire. Figuring out design flaws, trying to trick it's ally recognition patterns, or exploiting bad programming would probably work the best. As for say, the US fighting, I'd guess they'd just go with raw firepower, because like I said earlier, any problem you can't solve with overwhelming firepower just means you haven't used enough.
Last point - do not underestimate weather. Seriously. Another thing military history has taught us is that every general who decided to ignore the stuff lost. Badly. You'd need specialized bots in every weather type and pattern.