No. Drones are actually doing this as we speak... it will be far easier for an AI in 200 years to understand what they should or not do.
No current or proposed model of armed drone is allowed to attack a target without a human giving the actual command, and this is unlikely to change for the simple reason that nobody wants to encourage the robot rebellion, or for moral reasons allow completely automated lethal action. Speak of things in which you have some grounding. Humans are also infinitely less likely to break down, and a good deal harder to stop working correctly.
There's the commander to do this. One person. I entertained the existence of ten people.
Task overload denies you. A commander is absorbed in processing data and making decisions on targeting and strategy; while he will undoubtedly give course orders, a second person at the helm is required for safety reasons, as there are things the commander may not see. Similarly the commander is unlikely to be alone for the same reason, so there will probably be a second officer in the loop since a human's ability to process multiple tasks is limited. And so on and so forth, the larger the ship, the more people are required simply to command it effectively.
No. Easily accomplished by an AI.
Indeed? Then why is it so easy to screw with their heads on this in many games? AIs short of strong lack the flexibility and adaptability to handle this correctly.
No. Drones may eventually be designed to do so. Not such a big deal, if you think we have 200 years to design them.
You are proposing the existence of true AI in a humanoid platform then? Short of a true artificial intelligence the ability to adapt to unpredictable scope and types of damage would be difficult, nearly impossible; similarly any proposed drone must be able to maneuver both in gravity and non-gravity environments, reasonably good at dealing with extremes of EM radiation from both damaged equipment and enemy fire, and able to manipulate a wide variety of tools and materials.
Even if you can manage all that, the odds are good trained humans will be cheaper and easier to replace.
No. Trivial problem really.
As above: Humans are also infinitely less likely to break down, and a good deal harder to stop working correctly. Why do we still have humans in nuclear power plants? Surely we can construct a foolproof safety system by your logic?
Well, no, we can't. And trained humans are cheaper than trying.