FRC AprilTag Vision Examples

Quick start for FRC teams learning AprilTag detection and robot pose estimation. Detects AprilTags, computes robot position, and drives to game targets.

Architecture

VisionContainer wraps three vision system implementations:

• ControllerVision - Free software based on WPILib, requires a camera, and a roboRIO or a simulation PC
• PhotonVision - Free software, requires a camera, roboRIO and a coprocessor or a simulation PC handles both roboRIO and coprocessor
• LimelightVision - Purchased commercial hardware and integrated software requires a roboRIO or a simulation PC

Example Commands

Four targeting strategies are implemented as commands:

1. Drive to Field Position - 2026 Paired HUB Tags

Drive to a 3D field pose relative to the left HUB tag.

2. Drive to Relative Position - Either Single Tag of a Pair

Drive to a 3D pose relative to either tag of a pair. Two implementations:

• PID-based: Three PID controllers with stub drivetrain
• PathPlanner: On-the-fly command (partial example untested in this package)

3. Drive to Single-Target Relative Position

Drive to position relative to target using Transform3d instead of field coordinates.

4. Turn and Drive to Distance

Turn to target and drive to specified distance using yaw/pitch relative to AprilTag.

Note

Different alignment command examples target AprilTag #12, 9-10 and 19-20 paired HUB tags, or any of the paired tags (FRC 2026 RbuilT). You'll need to write your own game-specific targeting code.

Setup

Read the comments in Robot.java and each vision system wrapper class for configuration details.

Required to run this project:

• To run simulation of a roboRIO, a DriverStation PC is needed.
• To run with a real roboRIO, a roboRIO v1 (or v2) is needed.
• This project runs as is with a provided crude drivetrain and gyro.
• To test PhotonVision, that needs to be downloaded to at least the DriverStation PC (or a co-processor may be used).
• To test LimelightVision, that needs to be purchased (beg, borrow, ...).

Required if advancing to competitive usage:

• Your drivetrain implementation
• Your implementation of where to drive to scoring positions For 3-D pose usage the following are strongly recommended for accuracy and continuous driving even if the AprilTag is not in view:
• Odometry
• Gyro
• Pose Estimator combining odometry, gyro, and AprilTag pose estimation

Defaults:

• AprilTag #12 (change in code or print tag #12)
• Paired HUB tags #9 and #10 work for the default fake gyro for Limelight to match the heading
• Xbox controller (easily changed to other inputs)

Warning

Tune PID and timeout parameters for your specific robot. Current values are for hand-driven camera testing.

Hardware Requirements

Minimum (simulation/testing):

• DriverStation PC
• Optional ControllerVision (included in this project) is required to test that function
• Optional PhotonVision software is required to test that function
• USB camera (~$50) for ControllerVision or PhotonVision
• Optional LimelightVision device is required to test that function
• A roboRIO can be used to avoid simulation and implement vision on a robot
• AprilTags #14, #9, #10

For competition:

• Accurate vision systems require calibrated cameras. PhotonVision and LimelightVision provide for calibration and PV can be used to calibrate cameras for ControllerVision
• roboRIO v1 for ControllerVision turn and drive to target (yaw/pitch)
• roboRIO v2 for ControllerVision 3-D pose usage
• A dedicated coprocessor for PhotonVision or LimelightVision is recommended instead of using ControllerVision software on the roboRIO

Note

Running vision processing directly on the roboRIO is discouraged. Substantial CPU and timing tradeoffs may make it unsuitable for competition.

Vision System Options

System	Cost	Hardware
ControllerVision	Free	USB camera only
PhotonVision	Free	USB camera + coprocessor for real mode
LimelightVision	Purchase required	All-in-one device

PhotonVision on a coprocessor offers good quality at low cost. LimelightVision is simpler to set up but more expensive. ControllerVision software is free and runs on your roboRIO; it may suffice for "zero budget" teams.

There are a few other viable Vision Systems.

Running

This project runs on a roboRIO v1 with all three vision systems.

Simulation of a roboRIO mode works with all three vision systems.

Note

LimelightVision in simulation requires connecting your PC and Limelight to the same network.