Depending on how much you care about firearm realities, there's a significant chance that an unstabilized autoloader will fail to cycle, either stovepiping (as when you "limp-wrist" a handgun) or failing to return to battery. The fact that your pilot will be making tests to attack with a very small dice pool probably duplicates this adequately (high chance of a jam) but it's worth thinking about.

cndblank, I agree that a tiny little version of the Smart Firing Platform is possible, but I think you're lowballing. You need a node and a Pilot to run on it, for one thing (standard commlink and program costs, I assume), and you'd need Gecko Grip on two sides (unless it's a custom mount for a specific gun). The weapon would have to be a Smartgun to begin with, and it would have to have Electronic Firing. We're assuming (and that's okay, but notable) that the little servo widget and the gecko grip are strong enough to handle the recoil (not just survive it, but keep the gun pointed even vaguely in the right direction); that is, minimally as strong as a human hand just to get RC 0. I really think the result will not be cheap enough to justify *not* using a Smart Firing Platform.

It seems like a sticky grenade is a better option, or just invent a smaller 'not-so-smart' firing platform (reduced cost by limiting range of motion, drone stats, gun size, etc.).