The short version
We started with a robot that would collect both balls and shoot each with its own shooter. We built and tested parts of it, and learned a lot about squeezing a ball and sorting by size. In November we made two big choices: focus on Pollen for the qualifier, and switch to a topspin shooter after our shots kept bouncing out of the hive. That new robot is CR1. We cleaned it up into CR1 v6, and a sponsor credit turned its main plates into aluminum. Next we build it and test it before the qualifier on December 20.
1Sep 24–25
Make the robot ourselves
- We saw
- Our laser cutter bed (305 × 381 mm) and printer beds (256 mm) are smaller than a robot side plate.
- We chose
- Build in-house from 1/4 in plywood and PLA anyway: split big plates with seams that step around holes, and use one bolt size (#8-32) for every wood joint.
- Why
- We can change a part the same day, and every student learns CAD, laser cutting and printing.
- Entries
- EN-01 · EN-03 · Story E
2Oct 4–7
Find the real cause before fixing the shooter
- We saw
- Our first pollen shooter barely shot: about 1–3 m/s, or the ball got stuck.
- We chose
- Go through the video frame by frame first. The wheel wasn't touching the ball. We made a rigid hood with a 63.5 mm gap (about 10% squeeze), instead of a heavier flywheel or faster gears.
- Why
- A heavier or faster wheel can't help if it never touches the ball.
- Entries
- EN-12 · Story B
3Oct 18–21
Sort the two balls by size, and test it on the bench
- We saw
- Pollen (71 mm) and Nectar (91 mm) come in through the same intake.
- We chose
- A 78 mm slot that only Pollen fits through. We built a bench kit before the robot: Pollen went into the slot 5 out of 5 times, Nectar got past it 6 out of 6.
- Why
- A cheap bench test found a problem (Nectar stuck on a lip) that a grip strip fixed in one afternoon.
- Entries
- EN-27 · EN-36 · Story A
4Oct 23–26
Copy what works
- We saw
- The strengthened pollen shooter v4 was built, bench tested and "worked great". The old nectar shooter design had the same no-squeeze problem.
- We chose
- Start the nectar shooter fresh as a copy of pollen v4, sized for the bigger ball.
- Why
- Starting from a design that already works was faster than patching the old one.
- Entries
- EN-39 · EN-41 · Story C
5Nov 1
Pollen only for the qualifier
- We saw
- We had no reliable way to drive Nectar into its shooter, and the drivetrain felt rough.
- We chose
- A Pollen-only robot for the first qualifier, with stronger end-tied drive pods. Nectar waits for later.
- Why
- One thing done well beats two things half done. It also freed space, weight and a motor.
- Entries
- EN-57 · EN-59 · EN-61
6Nov 2–3
A short, full-width wedge for the flower
- We saw
- In video of another team, a wedge slid under the Pollen in the flower so it rolled out. Our first linkage reached 703 mm when deployed.
- We chose
- One full-width wedge on two short arms (at most 587 mm out in CAD).
- Why
- Fewer moving parts and it stays inside the size limit when deployed.
- Entries
- EN-63 · EN-72
7Nov 4–11
Switch to a topspin shooter: CR1
- We saw
- In class footage, the shots that reached the hive bounced back out, 3 of 3 that we could follow. Our hood shooter gives the ball backspin.
- We chose
- A new robot, CR1, with the shooter wheel on top of the ball and a grippy ramp under it. We kept the drivetrain, odometry, hubs and wedge, and added a turret.
- Why
- Topspin makes the ball roll forward into the hive after it lands. Another team found the same when prototyping.
- Entries
- EN-74 · EN-90 · CR1
8Nov 13–16
Test rigs before the robot
- We saw
- CR1 had four new parts we had never built: the topspin head, the turret, the lane and transfer, and a new wedge pivot.
- We chose
- A small bench rig for each one, made from the real robot parts. The topspin head shot much more consistently on its rig, so we made it one printed piece with the motor under the wheel (9 parts to 1).
- Why
- A rig is cheap to change. Mistakes found there don't cost a whole robot.
- Entries
- EN-91 · EN-93 · EN-94 · EN-100 · EN-102
9Nov 16–17
CR1 v6: fewer parts and a quarter inch to spare
- We saw
- CR1 had many one-off brackets and signs made of 16 parts, and only 0.35 mm of width to spare under 18 in.
- We chose
- A six-step clean-up: one bracket shape, one-piece signs, one chain idler, side plates cut whole, and the drive pods moved 5 mm in for 0.25 in on each side.
- Why
- A slightly proud bolt head could fail inspection. Fewer parts are faster to build and fix.
- Entries
- EN-104 – EN-108 · CR1 v6
10Nov 18
Keep the wedge arms outside
- We saw
- Wedge arms outside the pods could be hit by other robots.
- We chose
- We fully modelled arms inside the pods, then kept them outside.
- Why
- Inside meant moving the wedge gears, a new arm channel and many changed parts. The outside design was already checked, and the qualifier was a month away. Not every idea we model gets picked.
- Entries
- EN-109
11Nov 19
Metal where it matters
- We saw
- A sponsor gave us a $500 credit for metal parts, but the whole robot in metal would cost more.
- We chose
- The 7 plates that carry the wheels, motors and hits (side plates, motor plates, rear panel) in 1/8 in aluminum, with every part staying in the same place.
- Why
- Aluminum doesn't crush under nuts or split at bearing holes. Keeping every part in place meant no new clashes to chase.
- Entries
- EN-111 · v6 METAL
12Nov 19–20
Keep every ball moving, and let the camera see
- We saw
- About 200 mm of the lane had nothing pushing a ball. The parked wedge blocked the camera at most turret angles.
- We chose
- A lane printed in 2 pieces (was 7 parts) with 3 small extra wheels, the camera on top of the shooter head, and 3 low wedge shapes to test on the real flower.
- Why
- A ball that stops in the lane is a lost shot. A camera that can always see the target helps the turret aim.
- Entries
- EN-112 · EN-113 · EN-114 · EN-115
What's next
CR1 v6 is fully designed and checked in CAD, but not built yet. Before the qualifier on Sunday, December 20 we will:
- Cut the aluminum and plywood plates and print the parts.
- Build the robot and measure it against the 18 in box.
- Test with counts: pickups out of 20, shots in out of 10, flower pickups out of 20.
- Tune the shooter speed control and the turret aim, and write the results in the notebook.