Built to score
from the far zone.
A vector swerve chassis and a three-motor auto-aiming turret — the first swerve drive raced in FTC, validated by a regional championship. Explore the real CAD in 3D.
The
machine.
19859 · 2025–2026 competition robot
Vector 3.0
swerve.
This is the first swerve drive raced in FTC — a genuine technical breakthrough. Modelled on the FRC 10086 second-generation module (some of our members also run FRC), we re-engineered it to fit FTC's rules: the steering motors became X25 servos, and gears gave way to belts to cut weight for this season's fast, high-contact game.
Four independently steered and driven modules give full holonomic motion — translate, rotate, and translate-while-rotating. Field-centric drive fuses IMU data so the driver never has to track which way the robot faces. We solved the hard parts — inverse kinematics, assembly precision, driver load — and proved it by winning the Chongqing regional.
Three-motor
turret.
Built for throughput and a short scoring window, the three-motor turret plus chassis-based auto-aim lets us fire before contact ever happens. Full-field localization means we ignore the Limelight's field-of-view limit — the moment we're in position, we shoot. Auto-aim locks within 0.3s and reaches target RPM within 0.4s.
To keep flywheel speed stable through rapid triple-shots, we embedded twelve 10 mm copper beads as an internal counterweight — adding inertia without changing the envelope. A Limelight backs up the chassis aim whenever odometry drift crosses a threshold, keeping aim stable all match.
Built to be unmistakable.
Every part earns its place.
A full rubber-band roller on 58 mm wheels with 700 mm nitrile bands, widened past the bumper. Mecanum side wheels sweep wall and corner artifacts to the centre; a steel plate guards it through contact.
Static was randomly disconnecting our control hub mid-match. Inspired by a Jingdezhen ceramics outreach trip, we machined ceramic shims that block static conduction without blocking signal — a perfect fusion of craft and engineering.
Two X25 25 kg servos gate the ball path into the turret, so the intake can't over-feed and jam the flywheel. A GoBilda RGB indicator makes robot state readable at a glance on the field.
We don't guess.
We measure.
Every change starts from a real problem or a clear performance target — never change for change's sake. Every test logs raw data and is repeated at least three times. Each iteration cycle is capped at 3–5 days to avoid over-design.
Chassis wheel
FRC-style wheel with servo steering at 1:1 for response. Wide tyre testing showed sluggish steering and weak grip.
Swapped to VEX traction wheels — a perfect balance of response and grip, with the swerve's mobility now obvious.
Cast our own studded silicone tyre: huge grip (push a same-mass robot at 3 m/s) — but the grip starved steering torque.
Removed the studs and tuned the tyre's Rho curve for a moderate contact patch — grip without the steering penalty.
Flywheel & intake
RPM dropped noticeably after each shot — the second and third balls drifted further out. Inertia was the culprit.
Embedded twelve 10 mm copper beads in the flex-wheel gaps via a 3D-printed insert. Triple-shot speed held stable — structure untouched.
Too narrow, couldn't grab corner artifacts, and elastic kickback bounced balls away — a low auto ceiling.
Widened intake with mecanum side guides for wall pickup and a soft cast-silicone centre roller — smooth, high-tolerance collection.
The robot, in the wild.
