We seeded number one at Lakeview (week 4) and we were honored to play with the WiredCats (5675) and I Am Robot (4810) and picked up our first blue banner. A combination of a reasonable schedule and a healthy robot throughout the entire tournament were the keys to our success.
Changes to the robot included a 100% engaged center auton which used the gyro to mark when the robot was climbing back onto the charge station and then used odometry to travel a given distance, then used gyro again to micro move to ensure a good balance. The trigger to stop moving was when the gyro started to go back down to level. We also doubled up the belting on the arm’s last 2 stages to reduce the stress each belt carried.
We were placed on the Aptiv field after the reshuffle and had the misfortune of getting a schedule that had a SOS score, according to statbotics.io, of .95. We were glad to be able to play with our friends 8608, the Alpha Bots, and 3603, Cyber Coyotes. Highlight of our matches was going up against the number 2 seed, 123, 2337, 33 and pulling off an upset.
Changes to the robot included moving from LabVIEW code to JAVA in order to implement path planner to get a 2 cone auton. JAVA team had been working on code since the beginning of the build season and it was now their turn. We struggled to trust it all tournament. In a practice match we had flipped on the charge station and were concerned there were inconsistencies in our pathing that would make us likely to flip again. The lack of an untethered practice field at MSC made it so that testing could only happen during matches and I was dead set against showing any potential alliance captain our “suspect” auton. In the end, when we had nothing to lose, we ran the 2 cone smooth side and it performed flawlessly on the pick up. Drop off was always throw of the dice.
With MSC behind us, we were convinced that a 3 cone auton was required to seed well and be picked early at Worlds. We focused on a 3 cone, ground auton inspired by 862, Lightning Robotics. We ran it in all 10 matches and I believe it picked up the 2 outside cones 8 out of 10 times. We delivered the first 2 cones 100% of those times and the 3rd was an easy push in during teleop.
Ironically, we were chosen by the #1 seed, 2075, Enigma Robotics and 973, Greybots, who wanted us to only run our center cross and balance. We had not run our center auton since States and neglected to implement it into JAVA and thus we switched back over to LabVIEW. The RIO 2 has a removeable micro SD card which houses all of the software and so switching from JAVA back to LabVIEW was as simple as switching cards.
Because of subtle changes in our drive motors, settings of the auton were not perfectly tuned. We failed to balance during the first match. Tuning the balance was fairly straight forward but we also were failing to deliver the preloaded cone to the top node. Small tweaks in wrist angle and gripper opening angle finally allowed us to consistently deliver the cone in Finals 1 on the Johnson field. Had we run our center auton even once during quals, we would have seen these issues and fixed them fairly quickly on the practice fields. Instead, it was a mad scramble in the pits behind the stands where the stakes could not be any higher. The programming and pit team performed well under pressure with excellent guidance from Greybots and Enigma.
We made it to Einsteins, which still seems like a dream. We are incredibly grateful to be able to play with and against robots of such high caliber. We are thankful that 2075 and 973 gave us the time to get our “stuff” in order and thankful to have as capable of a team as 4607, CIS, to be subb’d in when we failed our climb on the charge station in Eisteins Match 1. We were told that the blue charge station on Johnson had caught a bunch of team in quals. Our lower bumper and chassis height made us susceptible for getting caught up. Unfortunately we would be blue and on the Johnson field for much of Eisteins. Even though we wouldn’t be back on the field, we are grateful to have stepped onto Einsteins at least once.
Keys to our success:
Chassis Design by Week 1:
We stepped into the season already with a solid chassis design that worked with SDS MK4i. We had inherited the design from 5205’s 2022 season. Our lead CAD mentor and our lead designer spent the week 1 adding the belly loaded battery mounting system. By the end of week 1, we were cutting pieces off our ShopSabre 48"x48" CNC router. By the end of week 2, we had a fully wired skateboard chassis that was ready for programming and drive practice. This early start in programming allowed us to add and tune auto turning algorithms in teleop and start playing with algorithms for the charge station balance. Without the CNC router, manufacturing would have been far slower. It probably would have taken 2X longer to get the chassis running and testing.
The chassis design was essentially a skateboard on top of which allowed us to add tooling. By separating chassis from tooling so distinctly, design teams for tooling were allowed to progress at a different pace than chassis. Chassis was able to be launched into manufacturing without any fear that it would need to be changed at a later date.
No Ground Intake:
The complexity of a ground intake would have forced us into far more prototyping and testing. As a rookie team with limited design experience, we made the early decision of not doing a complicated ground intake or a secondary handoff. This allowed us to focus on a simple arm design and an even simpler bucket design.
LabVIEW and then JAVA:
Just as we had inherited a chassis design from 5205, we also inherited LabVIEW code. I also had a highly competent sophomore LabVIEW programmer. She improved the teleop performance, added the arm code, and added a fairly straight forward auton routine. By sticking with LabVIEW, our JAVA team had an additional 6 weeks of time to get complicated code such as path planner and vision debugged. Getting LabVIEW done early also allowed us to transfer all of the arm settings directly into JAVA without much testing or tuning. LabVIEW’s superior debugging environment allowed us to prove out hardware faster and made transitioning to JAVA smoother.
Appreciate everyone who stopped by our pit at Worlds to say hi and learn about our robot. We are going to clean up our code and post it here in the coming weeks. CAD hasn’t really changed since we posted it here originally.
As always, we welcome all questions and comments.