Everyone else is more knowledgeable than me, and they have already answered your questions well. I want to add a few points, however:
- Let strategy dictate design, not the other way around. Know your priorities for the most important things to make the robot do, and incorporate them into the design first. You can add other cool features later.
- CAD everything IN A TIMELY MANNER. This means lots of communication between CAD and Build. We wasted time while each group waited for dimensions for parts from the other. They need to figure it out. That will cut out probably three to five days of doing nothing for many people.
- Think about making Electrical and Programming one, and definitely have collaboration at least. On a team of 25, we had seven people between those two groups. At no point did we ever need more than three (maybe four) working on wiring or programming something. A small group may mesh better, anyway.