Say, you've got a nuclear reactor (or whatever else is more practical) on an asteroid. You cannot use the energy from the reactor to propel the rock, you need to generate thrust. So you take water, you split hydrogen & oxygen, you burn it, you get thrust.
Oh, and fun safety fact: a nuclear reactor isn't actually particularly radioactive until it's turned on. If you launch it out of Earth's orbit before turning it on, you're sitting pretty.
Also, nuclear reactors aren't the ideal way to make electricity in space. Proper nuclear reactors work as heaters for a heat engine, and heat sinks are proportionally much more expensive in space than on earth.
However, solar power is much more efficient. You get 100% efficiency all the time, and it's 30% better in space (near Earth) than it is down here at best. Also, since no structure has to carry any significant weight, you can build things like a flexible paper-thin sheet of solar cells printed on plastic that is spun about it's axis to keep it deployed and spread out.
That energy could come from solar panels as you mentioned, with the drawbacks you mentioned. But it looks to be a good enough method.
I have no idea if that's the best way to convert solar energy to thrust, or if it's even feasible, but bringing perpetual motion into it seems a little harsh.
So you have energy collected over a long period of time, then released in a short period of time. That means you need to store that energy. It's not feasible to ship batteries up from the surface with current battery technology, since the energy/mass ratio is awful, so you'll need some other way to store the energy. Chemical storage would be one way. Use local materials and transform them into new materials with more chemical potential energy. Chemical storage tanks can be reasonably light. When the time comes, you can reverse the process and transform your stored chemicals back into electricity.
But wait, we don't actually want electricity, we want thrust. A traditional rocket engine is a great way of transforming chemical potential energy into thrust. So let's skip the electrical middleman, and burn our chemical energy storage reagents in a rocket engine.
All that's left to do is choose the chemicals. If there's water, then it's easy to make oxygen and hydrogen from it using electricity, and hydrogen is a great rocket fuel.
I still think it's very practical to consider moving asteroids around from a technical perspective. The political ramifications are huge. If the people in charge are malicious they could completely devastate the planet.
I think this is generally solvable by using lots of small thrusters that are independently controlled to make sure that no one can smash the thing into the planet, but getting sign-off from everyone who might have veto power is huge.
"PM: You've suggested an asteroid could be brought closer to the Earth to make it easier to mine. Is that really feasible? EA: It is. One of the ways that we could do that is simply to turn the water on an asteroid into rocket fuel and burn it in a thruster that nudges its trajectory. Split water into hydrogen and oxygen, and you get the same fuels that launch space shuttles. Some asteroids are 20 percent water, and that amount would let you move the thing anywhere in the solar system."
Wtf? EA just suggested a perpetual motion machine. He suggests using energy to hydrolyze water, but then de-hydrolyzing the water to get energy. He's making it sound like more energy will come out than went in (allowing propulsion), when really it would be == at best, and really < because of inefficiency. The only way his concept of storing the energy is even slightly useful is if you were to spend years and years collecting solar energy and storing it as H2 + O2 to burn later in one quick burst. But that's you storing the energy over a long long time, and getting back less than you put in.