n e x t t i m e ? ?
If we could do it again, we would have focused on a simpler project with less mechanical parts and less $$ to spend. We would have figured out a way to control the joints with less microcontroller output pins (there's got to be a CMOS chip out there for that!). We would have talked to more people who had done things like this.
It would have been nice to fully integrate the object recognition code- how fancy would that have been for the arm to detect and actually pick up things on its own? I think this was our "ultimate goal" for the project, and it was very tough to reach. However, we had fun experimenting with the GUI and it was pretty satisfying enough to make the arm move 4 different joints in sequence to a specified angle.