Help us create the perfect combat robot.

The ideal combat robot is, in my opinion, too broad a topic for a single dissertation. I'm no specialist, of course, but this probably requires at least 3-4 people working on it. One dealing with mechanics and/or electric drives, a second with electronics, a third with programming... Someone else should be managing all of this; basically, this is a task for an entire design bureau. By the way, no torsion fields have been detected yet—durkalurka guarantees it. Most likely, they don't exist at all.

Added after 3 minutes
For a dissertation, it would probably be enough to even just teach it to distinguish between friend and foe with a probability close to 100%..