[Al Skierkiewicz]

Chris A did a very good job of explaining "Crab" steering. Simply put the mechanism mimics the way a crab moves across the sand of a beach. It literally takes off in any direction with it's eyes always fixed at the object it is running toward or away from. This is a term that is also used in broadcasting as it is the way a camera dolly is steered to move the camera across the studio floor with the lens always pointed at the talent.
Our first attempt at crab used two drive wheels coupled by chain and driven by a motor. The "front" wheels in that robot where casters just to follow the power delivered by the drive modules. We used this design in the year of floppies and puck.
This year we had independent front and rear coupled pairs. Each pair had a steering motor and pot and were couple from side to side by chain. The advantage to this steering is there is no need to turn the robot to drive in a particular direction.
As Chris stated above, there is a need for feedback and that comes in the form of a pot attached to one of the steering axis to provide a reference for which direction the module is turned to. Rotation is limited to less than 270 degrees (the max rotation of a pot) to simplify design. This year software design allowed the robot to steer using crab, normal two wheel steer but four wheel drive, or full up tank drive and since the front and rear pairs were independent, the front could be turned left while the rears were turned right providing a distinct center of rotation for controlling a goal in a struggle with another robot. With practice we were able to transfer the center of rotation from inside the robot to the center of a goal or the virtual center of two goals. (yes, we were able to grab and control two goals.)