Our robot was designed to ridiculously exacting requirements, and was intended to meet the design specifications to the letter. For no apparent reason, other than its sheer impressiveness and the prospect of avoiding the need to drive, we elected to create a non-moving, catapult launching robot which would nail the 3 pointers from downtown.
The actual robot was fully, and I mean FULLY, designed and assembled in SolidWorks prior to cutting. This was an immense help in determining space availability, clearance problems, visualization, and ultimately allowed us to very accurately cut all our parts on the lasercamm without the need to hand drill or cut any parts. Also, it allowed us to use every ounce of available space and to waste nothing. Also, the Great White Hope is gorgeous.
Simple, really: if it gets put in the corner, it had better be circular to spin around anyhow. Eventually extra layers would be added to the robot to provide simple space efficiency. Layers were mounted on ¼-20 rod to ensure simple disassembly if needed and the ability to finely adjust component heights and clearances. This proved to be exceedingly useful in the mechanical debugging stages. The whole this was mounted on a nice, solid aluminum block through a ball bearing turntable to give us that lovely rotational ease of motion.
The catapult was intended to be super accurate by allowing fine adjustments via a stepped wheel driving the edge of a circular platform to give us an angular resolution of less than a ¼ of a degree. This platform was mounted around its axis on a 1/4 -28 screw with a small, but smooth, thrust bearing sandwiched in between the main chassis plate and the rotating catapult stage. In order to allow its drive wheel to be mounted next to the low slung platform, the stepper motor was hung from a small plate suspended on two of the four ¼-20 body rods.
The catapult was a spoon clamped to the motor via a small machined mount which grabbed onto any variety of spoons you gave it using an ingenious set screw. This allowed infinite variation and fine-tuning which we never really need that badly. However, it was cool. The mount was slipped and glued directly to the drive motor shaft which was, in turn, mounted as high above the platform as possible (with constricting clearances of surrounding components) to allow maximum motor spin up time. If it was too low we wouldn't build up enough speed to fling those balls to the baskets. Again everything was cut to spec sheet tolerances on the Lasercamm, making assembly extremely easy.
The ball loaders were designed to mimic what we, and apparently several other groups who copied us, thought was the best system of yester year - the six-shooter barrel loading tube design. We found, via Solid Works, that three separate 5 tube ball loaders with 4 balls worth of space per tube gave us the most room for the other necessary components while giving us great capacity in a compact space (61 balls if you include the pre-loaded catapult).
We did, unfortunately, find that using 3 separate systems makes the "interference plate": which prevents all the balls in each tube from falling out at once very hard to position properly. It just kept banging into the tubes and stalling the motors. Luckily we never got to compete and we just took it off for checkoff, since our shooter was great and could have gotten along fine with 3 balls, let alone the 14 left on the first level to start.
The barrels were held in place by 5 sided "barrel stars" which were cut to fit the tubes into position. These were then bolted to ½-13 rods which were drilled axially to fit directly on to the stepper motor drive shafts. This worked great, but twice the rods needed to be reglued since the torque loads got too high when they seized during the debugging stages.
The best part may have been the small ramps which deposited the balls from the drop points to the catapult spoon. The spoon head was placed exactly at the pivot point of the catapult platform, so it always remained in the same spot. This allowed us to run the ramps directly to the spoon creating what was effectively a funnel to the catapult. If you were to have thrown a ball into the loading area it would quickly settle into the same spot on the spoon every time. Every time. Oh yeah.
We ditched the idea of using a centrally mounted stepper motor to spin the platform when we had an epiphany. One wheel, in the back, recessed to fit inside the circle. Simple, easy, reliable. The only potential problem: the weight of the robot is on the bearing turntable, not the wheel(s), so it has little contact force if improperly placed. You might, perhaps, have to file down the slots in your motor mounting plate which you bolt to the chassis with if the wheel doesn't make sufficient contact with the ground since everything in your solid CAD model was set to be perfectly tangent to surfaces and not to apply any contact forces. But that's purely hypothetical. No one would ever overlook such a problem. It might arise in several places where round things touched, and that would be silly. Very silly.
A beautiful, fully mechanically functional robot, right out of the proverbial lasercamm box. Seriously, it was a thing to behold, newly assembled and unscathed by failure or frantic tools in tired and weary hands. I loved it. We all did. But then things got sticky, and our beautiful baby got finicky. Upon the addition of electronics we found ourselves with a robotic trophy wife. Pretty, but it didn't do a thing. Thus, we move on to electronics.