Tuning Workflow Overview
Limelight's own documentation has all of the tuning guidance in this section, but it's spread across a dozen unrelated pages: sensor settings live under "Getting Started," color thresholding lives under the retro pipeline, MegaTag2 IMU tuning lives under AprilTag localization, and PID starting points are buried in a tutorial. This section reorganizes the same guidance into the order you'd actually tune a robot, cross-linked to the YALL API where a step is configured from robot code instead of the web UI.
The order that works
Camera sensor settings: exposure, black level, gain. Do this first, regardless of pipeline type. Every later step (thresholding, detection, pose accuracy) is easier against a clean image.
Crosshair calibration: only if you're driving off
tx/tydirectly (visual servoing) rather than a full 3D pose.ChArUco camera calibration: only if you need accurate 3D measurements: AprilTag 3D tracking, robot localization, or
getXXXPose_YYYSpace()calls in general.Pipeline-specific tuning: pick the one matching your active pipeline:
Retroreflective pipeline tuning: HSV thresholding, then contour filtering/sorting.
AprilTag & MegaTag2 localization tuning: robot-space offset, field map, IMU mode, assist alpha.
Neural network pipeline tuning: confidence threshold, crop window.
PID tuning for aiming & ranging: once target tracking itself is stable, tune the control loop that drives on it.
See Also
Last updated
Was this helpful?