The 4 laws of manipulators:

1. Let strategy drive your manipulator design, not vice versa. What this means is that your manipulator should achieve the design requirements your team decided upon in forming a strategy, instead of forming a strategy based upon the manipulator you cobble together.
2. Design within your teams’ limits. If you know you just don’t have the resources to build the super-complicated manipulator that you think will dominate every aspect of the game, don’t do it! Go for the simpler one that you can build and will fulfill one role really well. However, also don’t be afraid to push your team to overcome your limits. For example, my team pushed ourselves to build a practice robot this past year, and it ended up being really beneficial.
3. Always have active control of the gamepiece. For example, if a ball needs to be transported through your robot, do it with an active conveyor, not a ramp. If you don’t actively control the gamepiece, it will inevitably jam or fall out of your manipulator. An experienced mentor from 1778 - Ian Curtis would always say “Gamepieces will break your heart.” Make sure that the gamepiece is always under active control.
4. Prototyping and iterative development are key to building a successful manipulator. Start out with a prototype, and then iteratively improve it until you are ready to build a final version. Even then, be looking for improvements that will make it better.

Based on : http://www.instructables.com/id/Introduction-to-Manipulators/ by first intern

More great information on design of a robot here: An Idiot’s complete guide to FRC