My team has used a single Limelight3 to get localization data off April tags the last couple years with moderate success. We’ve always come to a stop, got a snapshot, and then adjusted our pose 100% based on said snapshot. We’ve done pretty good, and I’m pretty sure our shortcomings have been more due to learning curve rather than hardware inconsistency. That being said, we want to take a leap this year and go from being “pretty good” to being “really good”. Should we focus more on using the WPIlib pose estimator (add vision offset) tools, or should we look more first at camera calibration. I fear diving into the calibration stuff may cause us more inaccuracies than what comes out of the box with a new camera. Any advice? Thanks!
The stock calibration with LL can be off by as much as 15%, so I’d definitely recommend you calibrate it ASAP.
1 Like
Especially important for the Limelight 3G, which comes unfocused. Make sure to focus the lens, glue it in place, and only then run the calibration.