Ben
Final compact FTC robot with bumper intake and odometry-controlled turret
Back to projects

FIRST Tech Challenge · 2025–2026

FTC Competition Robot

Three robot iterations led to this compact turreted design for the NYC Championship.

Season

Three complete robot iterations

Final robot

Bumper intake + servo turret

Season finale

NYC Championship

The challenge

FTC DECODE game requirements

In the 2025–2026 game, robots collect purple and green ARTIFACTS, launch them into a GOAL, and use their colors to complete patterns.

Official DECODE overview ↗
CAD of the final FTC robot with bumper intake and turreted shooter

Game pieces

Purple and green ARTIFACTS

Main task

Collect, aim, and launch into the GOAL

Size constraint

Fit inside an 18 × 18 inch footprint

Shared shooter control

Flywheel RPM control

Every robot used the same core shooting strategy. Field position determined the distance to the GOAL, testing converted that distance into a target flywheel RPM, and closed-loop control kept the flywheel near that speed.

01 · Measure

Distance calculation

Odometry tracked the robot's field position. The distance formula compared that pose with the fixed GOAL coordinates.

02 · Select

Distance-to-RPM regression

Repeated shooting tests produced a regression that mapped distance to the flywheel speed needed for a repeatable shot.

03 · Correct

PID speed correction

Encoder feedback measured actual RPM. PID continuously adjusted motor power to reduce error and recover after each launch.

Measured flywheel response

RPM recovery after a shot

FTC Dashboard graph showing flywheel RPM dropping after a shot and recovering toward its target
The target came from the tested distance regression. The RPM dip shows energy leaving the flywheel; the recovery shows the PID controller driving it back toward the requested speed.

Closed-loop flywheel control

Flywheel PID feedback loop

Flywheel PID feedback loop Target RPM is compared with measured RPM. A PID controller changes motor power and encoder feedback closes the loop. TARGETRPMΣ+−CONTROLLERP + I + Dcorrect RPM errorMOTOR POWERFLYWHEELENCODERmeasured RPM same control strategy on both robot versions
The controller continuously compared requested and measured flywheel RPM. Version 2 kept this loop, but two shooter motors recovered to the target speed much faster between shots.

Robot version 01

Version 1 mechanism layout

The first robot tried to create one continuous path from the floor to the GOAL. It combined a rubber-band intake, sideways centering, a kicker, a flywheel shooter, and an AprilTag-guided turret.

Collect

Rubber-band intake

Pulled ARTIFACTS in from the floor

→

Center

Vectored wheels

Moved each ball sideways into the firing path

→

Lift

Servo kicker

Tried to push the ball upward

→

Launch

Flywheel + hood

Single-motor wheel with an adjustable exit angle

→

Aim

Camera turret

Arducam, AprilTags, motor, and gear ring

Version one FTC robot on the DECODE field surrounded by purple and green artifacts
Version one on the field with the complete intake-to-shooter path.
Close-up of the FTC robot turret, shooter housing, gear ring, camera, and wiring
Turret gearing, shooter housing, camera, and cable routing.

What testing exposed

01

Problem

The rubber bands stretched, tangled, and failed at useful intake speed.

Response

We temporarily installed shorter bands, then treated the intake material and geometry as a redesign priority for the next robot.

02

Problem

The vectored wheels did not compress the ARTIFACT enough to center it quickly.

Response

The slow sideways transfer showed that passive wheel angle was not enough without controlled compression.

03

Problem

The servo kicker could not give the ball enough energy to reach the main flywheel.

Response

We replaced it with a motor-driven feeder wheel that could shoot the ball upward into the main shooter.

04

Problem

The rotating turret cable could extend beyond the legal 18 × 18 inch footprint.

Response

We zip-tied the cable down as a temporary competition fix and carried cable management into the next design.

FTC team members repairing and rewiring version one of the robot
Between-event fixes included shorter intake bands and restrained cable routing.
Side CAD measurement showing version one's mechanism extending beyond the robot frame
CAD measurement used to check how the mechanism and cable path related to the legal footprint.
CAD of the original single-motor FTC shooter
Version 1 used one shooter motor. It ran the shared regression-and-PID control strategy explained above; Version 2 later added a second motor for faster spin-up and recovery.

Turret test

AprilTag-guided rotation

The front Arducam scanned for AprilTags, and a geared motor rotated the shooter turret to face the target. The same rotating assembly also created a cable-management problem that pushed the robot outside its legal footprint.

Robot version 02

Version 2: floating intake and chassis aiming

The second robot addressed the first intake's reliability and compression problems. It kept the same basic shooting idea, but doubled the shooter motors and temporarily traded the turret for field-oriented chassis aiming.

A time crunch shaped this version: removing the turret simplified the build, but moved a much larger part of the aiming problem into code.

Collect

Direct-drive rubber wheels

Pulled ARTIFACTS in without stretchable bands

→

Guide

Vectored wheels + funnel

Centered off-axis pickups toward the shooter path

→

Buffer

Trapdoor

Stopped the direct intake from feeding too early

→

Launch

Dual-motor flywheel

Smaller hood with faster RPM recovery

→

Aim

Rotate the chassis

Odometry, trigonometry, regression, and Limelight

CAD view of version two showing the direct-drive intake, vectored wheels, funnel, trapdoor, and shooter
Direct-drive rubber wheels pulled ARTIFACTS into vectored wheels and a funnel. A trapdoor held them back from the shooter until it was time to fire.
Front view of version two at competition showing its funnel, intake wheels, and Limelight
The Version 2 front end at competition, with the Limelight mounted above the direct intake path.

Problem

A rigid intake made the chassis bounce over the ball.

Floating response

The vectored-wheel assembly could rise through a limited range.

Remaining issue

The intake still could not reliably reach the funnel's outer edges.

Version 1 → Version 2

System

Intake

Changed

Rubber bands became direct-drive rubber wheels.

Why

The first bands stretched, tangled, and failed at useful speeds.

Result

A more positive intake path without elastic bands.

System

Compliance

Changed

The vectored-wheel assembly gained limited upward travel.

Why

A rigid intake transferred the ball impact into the chassis and made the robot bounce.

Result

The intake rose over the ARTIFACT while the chassis stayed planted.

System

Shooter

Changed

A smaller adjustable hood and a second flywheel motor.

Why

The shooter needed to reach and recover its commanded RPM faster.

Result

Much faster spin-up while keeping PID speed control.

System

Aiming

Changed

The turret was temporarily removed and the whole chassis aimed instead.

Why

The team was time-crunched and could not finish the turret for this version.

Result

Odometry and field geometry calculated the heading and shot speed.

Field-oriented targeting

Pose → goal vector → required heading

Engineering reference showing robot pose, target position, distance vector, and atan2 heading calculation
This reference visualizes the same pose and atan2 math. Although the graphic labels a turret servo, Version 2 used the calculated heading to rotate the entire chassis.

Vision

Limelight AprilTag detection

Version 2 moved from the Arducam to a Limelight for higher-frame-rate AprilTag scanning. Vision supplemented the robot pose used by the field-oriented targeting code.

Autonomous

Road Runner trajectories

By this version, the robot also had an autonomous routine built with Road Runner for the autonomous period of each match.

What still failed

Limited intake coverage at the funnel edges

The funnel helped guide off-center ARTIFACTS, but the outer sections extended beyond where the intake wheels could reliably grab them. Compliance solved the bounce problem without fully solving pickup coverage.

Rapid-fire test

Three-ARTIFACT rapid-fire test

This portrait test clip belongs here because it shows the combined result of the trapdoor, direct feed, dual-motor flywheel, and PID recovery between shots.

SITHS Mercury FTC team together after finishing as Super Qualifier finalists
The team after reaching the Super Qualifier final.

Competition result

2nd

Super Qualifier finalist

This version carried us through the Super Qualifier bracket and into the final, where we finished in second place.

SITHS Mercury engineering and outreach presentation board featuring the FTC robot
The robot documented on our engineering and outreach display.

Robot version 03 · Final

Version 3: bumper intake and servo turret

Our final robot kept Version 2's floating direct intake, trapdoor, dual-motor shooter, and odometry. It replaced the funnel with bumpers, returned to a turret, and compressed the drivetrain packaging for the robot we brought to the NYC Championship.

This was our most complete design of the season. We performed well at the championship, but our season ended there rather than advancing to Worlds.

Collect

Floating rubber-wheel intake

Kept the direct-drive intake and controlled compliance

→

Guide

Bumpers + vectored wheels

Deflected impacts without pushing ARTIFACTS away

→

Buffer

Trapdoor

Retained ARTIFACTS until the shooter was ready

→

Launch

Dual-motor flywheel

Kept fast spin-up, PID control, and adjustable exit angle

→

Aim

Odometry turret

Servo-driven 1:1 modular gears with no camera

Overhead field view of the final robot showing its compact chassis and rotating shooter
Moving the drivetrain motors toward the rear reduced the overall packaging around the intake and rotating shooter.
Close-up of the final robot's shooter turret, servo, and modular gears
A servo drove modular gears at a 1:1 ratio to rotate the complete shooter assembly.

Version 2 → Version 3

System

Ball guidance

Changed

The funnel became low-profile bumpers around the vectored wheels.

Why

Ball impacts against the funnel could push ARTIFACTS away before the wheels established contact.

Result

The bumpers allowed a smoother path into the same vectoring mechanism.

System

Aiming

Changed

A servo-powered turret returned with modular gears at a 1:1 ratio.

Why

Rotating the whole chassis limited driving and shooting to the same heading.

Result

The shooter could track the GOAL while the drivetrain faced or moved elsewhere.

System

Localization

Changed

The turret relied on odometry without a dedicated aiming camera.

Why

The field pose already supplied the geometry needed to calculate the GOAL angle.

Result

The Version 2 targeting math moved from chassis control to turret control.

System

Packaging

Changed

The drivetrain motors moved toward the rear of the chassis.

Why

The final robot needed less crowded space around the intake and rotating assembly.

Result

A smaller overall robot with cleaner mechanism placement.

Camera-free targeting

Odometry-based turret targeting

The robot already knew its field pose and the GOAL's fixed coordinates. The same trigonometry from Version 2 now produced a turret angle instead of a chassis heading, so the drivetrain could keep moving independently of the shot direction.

NYC Championship

Championship result

Version 3 combined the best working ideas from the season and gave us independent intake, driving, aiming, and shooting at the championship level.

Season takeaway

Design changes across three robot versions

The final design was not one sudden breakthrough. It combined the intake compliance, trapdoor, flywheel control, odometry, and packaging lessons earned across the first two robots.