2025–26 DECODE season robots

Two robots, one season.

Elite Ball Knowledge carried us through States. Then we tore it down and built the Worlds bot around one question: how fast can one artifact get from the floor to the goal?

CAD render of Elite Ball Knowledge, our States robot

States robot · 2025–26 DECODE season

Elite Ball Knowledge

Eight motors and six servos around a Geneva colour sorter. We modelled every part in CAD before anything was cut, then made it in the NCSSM FabLab.

Full CAD model of Elite Ball Knowledge
Motors
8
Servos
6
RPM mecanum drive
435
3D printed parts
100+
CNC parts
16
Shooter cycle
1.7s
CAD render of the shooter
The real shooter: orange hood above the intake on our robot

Shooter

Adjustable hood and weighted flywheel on a belt-driven turret that aims itself.

CAD render, then the real part.

CAD render of the spice rack

Spice Rack

Sorts artifacts by colour using a Geneva mechanism with a powered stick transfer.

CAD render of the intake
The real intake: a green artifact at the pink rollers

Intake

Two-layer compliant roller intake feeding a steep ramp straight into the rack.

CAD render, then the real part.

CAD render of the drivetrain

Drivetrain

Mecanum belt drive with dual odometry pods and a PTO that lifts the robot.

Nothing worked the first time.

The intake and transfer each took three versions and the Spice Rack took two. Here's what broke each time and what we changed.

Intake

Getting an artifact off the floor and into a sorting slot without stalling or squashing it.

  1. Intake V1: Active intake
    V1

    Active intake

    Faster than a four-bar claw, but bootleg wheels caught on the artifact notches and the ramp slope pressed balls flat.

  2. Intake V2: Compliant rollers
    V2

    Compliant rollers

    Silicone rollers and a chain-driven compliant bar. A steeper ramp with guides aims the artifact at the slot.

  3. Intake V3: Two-layer compliant
    V3

    Two-layer compliant

    A second powered row with surgical tubing closes the gap above the rollers, so artifacts stop slipping and breaking colour detection.

Spice Rack

The sorter has to stop dead every time the transfer fires — and a gear-driven spindexer never stopped in the same place twice.

  1. Spice Rack V1: Simple spindexer
    V1

    Simple spindexer

    Minimal artifact support. Hard to start and stop cleanly; ended up a centimetre or two out of position and bounced balls back.

  2. Spice Rack V2: Geneva mechanism
    V2

    Geneva mechanism

    A Geneva gear lets the motor run continuously while indexing to six fixed stops — under a millimetre of error, tuned to the intake and shooter.

Transfer

Moving a sorted artifact into the shooter fast enough to keep up with the Geneva.

  1. Transfer V1: Stick transfer
    V1

    Stick transfer

    Carried over from our Robot in 5 Weeks bot. Pushed the ball up mechanically, but reset too slowly to use the Geneva's full speed.

  2. Transfer V2: Servo wheel
    V2

    Servo wheel

    An Axon servo driving omni wheels. Right idea, not enough power — balls stalled in the dead space and stopped the rack.

  3. Transfer V3: 6000 RPM roller
    V3

    6000 RPM roller

    Motor swap plus a third roller to cover the dead space. Sorted balls now shoot in tandem — 2–5 second cycles including sorting.

How we chose materials

We scored each way of making a part on nine criteria, including strength, weight, repairability and cost. 3D printing scored highest overall, so it makes the non-structural parts. goBILDA covers standard hardware, and we save the waterjet for the drivetrain, which takes hits all match.

  • 3D prints56.5

    Fast, cheap, light, repairable

  • goBILDA46.5

    Strong and precise, always in stock

  • Waterjet32.5

    Reserved for high-load structure

Worlds robot · 2025–26 DECODE season

We tore it down and started again.

After the State Championship we stopped iterating and started over. We built the Worlds robot with our sister team SigmaCorns #22377, and it trades the spindexer and the lift for a single straight-shoot path. Cycle time dropped from 8 seconds to 2.5.

CAD render of the Worlds robot
Cycle time, down from 8s
2.5s
Scoring rate vs. States
2×
Artifacts held
3
Shoot-on-the-move turret
360°
CAD render: vectoring intake

Vectoring intake

Curved guides jammed on multi-artifact pickups. V3 uses right-biased vectoring wheels and a shallower compliant ramp, with dead wheels at each end that let artifacts go while the intake keeps running.

Finite element stress map of the drivetrain sideplate, blue for low stress through red for high

FEA-optimised chassis

Finite element analysis on the drivetrain plates showed us where material was doing nothing, so we pocketed it out. The chassis is 15.2″ × 14.5″, with the battery and four motors low in the base.

CAD render: ranking-point strategy

Ranking-point strategy

We scored each ranking point against its mechanical cost and committed to the goal RP. That let us cut the lift and the brake from the robot.