This is a long story, but one that shows the steps of continuous improvement.
Build & Week 1:
To add a little to what apalrd already said, we tried about 7 collector concepts. We made no fewer than 4 “final” systems. We initially didn’t believe the over under pinchers would be legal. We built a vertical set (similar to 71), but felt that it had too limited a range for a 6x6 drive system. We had several other concepts that looked really promising, but once applied to the robot were quite frankly failures. Our vacum required too much driver precision, and our center feeding belt system (while excellent in prototype phase) bounced the balls off of itself during the implementation phase. At that point we went to a high speed wide overhead roller with a lower reaction roller that allowed for the ball to reverse idle. This took several days of tuning to get it to collect at high speed without bouncing the ball out. In the end, it worked pretty well, but the ball would occasionally cam out during turning maneuvers, and would often pop free when hit by another robot. Our team lead had been pleading for us to try an over under system, but that would require a floating mechanism (8" wheels) I didn’t think we could get done in time. Moving over the bump smoothly was a key component of our strategy and a lower reaction bar really needed 4" traction wheels. Then came the 148 video. I didn’t sleep that night. I may have posted it as “nightmares”, but they were nightmares of the best kind. The next night (two days before ship-date), the kids and I made the first version of our pincher collector. With only 1 days worth of development, and Kettering being week 1, we decided not to chance it, and shipped with the overhead roller. At Kettering there were 3 teams with pincher style of collectors. Initially 2 of the 3 fumbled the ball more than our backspin roller (one is now a multi-time national champion). Then they got them tuned in, and they were amazing. Each one of the 3 got at least 1 penalty that weekend for an inadvertant carry, but it was clear that was a better solution. The backspin would have worked much better if our spin center was about our nose (think 4x4 with grippys on one end and slicks or omnis on the back). Unfortunately, our CG ended up 2 inches rearward of the middle wheels instead of forward. If we had 4 more pounds, we could have changed this, but we were at 119.8 lbs.
Its a funny thing having the robot in a bag for 3 weeks. It is agonizing not being able to tinker with it, but so much fun to plan the time that you will get with it. With the 6 hours of unbag time, we made some improvements and installed our pincher grabber. We had very little time to develop it, but it looked promising. Troy (week 4) was the first competition that the pincher premeired at. There it was a Ref magnet, and a penalty collector. Due to the nature of our backstop, the ball could wedge itself against the backstop and manage to lift as much as 1" off the ground even with 10+ pounds of downbungee on the collector. One match we had soo much downforce, the floating assembly actually lifted the front wheels off the ground, and we could no-longer traverse the bump. We weren’t the only ones getting penalties that weekend for carries (the reffing was quite diligent), but we did get more than most. With 12 matches, there are a lot of opportunities for experimentation. After 8 matches, we were basically penalty free, and the mechanism was way better than our backspin roller. It is what allowed us to take on that infamous alliance in the finals at Troy.
Week 5 MSC:
For MSC, we found another lb or two and switched the back wheels to omnis to add some maneuverability. this change allowed us to remove most of the rock in our system ans still turn well. MSC is also where apalrd’s software was fully implemented. We had been looking for a clutch mechanism, but couldn’t find what we wanted so Andrew fixed it in software (chaning some 0 to 1s was a lot lighter and cheaper than most of the clutches I found). I think we recieved one penalty and it was after being double-teamed and knocked around. The reffing was a little more lax as twice, I saw us grab a ball as we were headed over the bump from the far zone. Both times the operator kicked it as soon as she saw it, but clearly the ball was lifted off the ground for an ever so brief moment.
For Atlanta, there really weren’t any changes. It seems like it takes 3 competitions for a Killer Bee to find its place. This is true since I have been on the team. If there were 2 regionals, then there would be changes for the Championship. If there were 3 regionals, then the Championship would mostly be refinement. This year was earlier calm. There were a couple software upgrades, but the biggest improvements were working with the comp team to set the robot up better for auto-mode, and discussions of which balls to leave alone on the field.
Continuous improvement turns good to great. Continuous tinkering can lead to inconsistent behavior and headaches.