Limelight

A Limelight is a dedicated vision system designed for robotics.

Instead of performing all computer vision processing on the Robot Controller, a Limelight has its own processor that analyzes camera images and sends the results to your robot.

The architecture looks like:

Camera
   ↓
Limelight
   ↓
Vision Processing
   ↓
Detection / Pose Data
   ↓
Robot Code

This makes the Limelight different from a typical VisionPortal setup, where the Robot Controller manages the camera and vision processing.

Why use a Limelight?

Computer vision can require significant processing power.

A dedicated vision device allows the robot controller to focus on running the rest of the robot program while the Limelight handles vision processing.

Limelight also provides a collection of prebuilt vision capabilities, which means you do not need to write every vision algorithm yourself.

Common applications include:

• AprilTag detection
• Robot localization
• Target tracking
• Color-based detection
• Object detection
• Neural network-based detection
• Custom vision pipelines

Pipelines

A Limelight uses pipelines to determine how camera images should be processed.

A pipeline defines what the Limelight should look for and how it should analyze the camera image.

For example, a color-based pipeline might look like:

Camera Image
     ↓
Color Filtering
     ↓
Object Detection
     ↓
Target Position

An AprilTag pipeline might instead look like:

Camera Image
     ↓
AprilTag Detection
     ↓
Tag ID + Pose

Pipelines can be configured and tested through the Limelight's interface.

AprilTags

Limelights can detect AprilTags and provide information about the detected tags.

Depending on the configuration, this can include:

• Tag ID
• Target position
• Target orientation
• Target area
• Image location
• Robot pose

This makes Limelight useful for the same types of applications discussed in the AprilTags page:

• Alignment
• Navigation
• Localization
• Autonomous routines

Pose estimation

A Limelight can also estimate the robot's pose on the field.

Conceptually:

AprilTag
   ↓
Camera detects tag
   ↓
Limelight estimates pose
   ↓
Camera position is transformed
   ↓
Robot field pose

This can provide another source of localization information alongside encoders and an IMU.

As with any vision-based pose estimate, accuracy depends on factors such as:

• Camera calibration
• Tag visibility
• Camera placement
• Lighting
• Viewing angle
• Distance from the tag

Communicating with robot code

The Limelight performs its vision processing independently from your robot program.

Your robot code communicates with the Limelight and reads the resulting measurements.

For example, your program might receive information such as:

Target detected: true
Horizontal error: -7.2°
Distance: 34.1 in

Your robot can then use those measurements to make decisions.

For example:

Vision Measurement
       ↓
Calculate Error
       ↓
Controller
       ↓
Motor Power
       ↓
Robot Movement

This becomes especially powerful when vision is combined with control theory.

For example, a robot could use the horizontal error from a Limelight as the input to a PID controller to automatically align itself with a target.

Limelight vs. VisionPortal

Both Limelight and VisionPortal can provide powerful vision capabilities, but they work differently.

Neither is automatically better.

VisionPortal is a natural choice when you want to stay within the FTC SDK and use its built-in vision processors.

A Limelight can be attractive when you want dedicated vision hardware and the additional tools provided by the Limelight ecosystem.

When should you use a Limelight?

A Limelight can make sense when:

• Your team needs dedicated vision processing
• You need advanced AprilTag or localization capabilities
• You want to reduce vision processing on the Robot Controller
• You want configurable vision pipelines
• You want to use the tools provided by the Limelight ecosystem

However, a Limelight also adds another piece of hardware and another system that your team needs to configure and maintain.

For a simple AprilTag detector, VisionPortal may be all you need.

Learn more

This page is intended to explain what a Limelight is and where it fits into an FTC robot, not replace the official documentation.

For hardware setup, pipeline configuration, APIs, AprilTags, localization, and advanced features, see the official Limelight documentation.