CD Programmers,
A lot of robots last season had multiple Limelights to filter out noise and increase targeting accuracy. Does anyone have any input on how much their swerve odometry was improved from fusion of wheel odometry & one limelight to two limelights?
Having more cameras means you can have an AprilTag in view more of the time, leading to higher accuracy localisation. More cameras is generally better, with diminishing marginal gains for each additional camera. I’d say 2 is definitely worth it.
I plan on using 3 cameras connected to a coprocessor running PhotonVision.
It’s important to note that 1 camera vs 2 cameras and 1 coprocessor vs 2 coprocessors are two different questions. The limelight makes this question moot, but using Orange Pi 5s and PhotonVision, we had a 4-camera, 2-coprocessor setup for about the price of one Limelight, and every one of those cameras gave us equivalent detection performance to a Limelight.
We went from 2 cameras crossing each other’s field of view in the front of our robot to 4 cameras all pointed outward. We noticed a significant increase in our ability to use the AprilTags from the midfield zone, mostly because the closest tags to us were more often in view.
It’s too bad its moot, it’d be nice to see if usb global shutter implementation and localization occur for the LL3s and 2s.
@Brandon_Hjelstrom any word on this feature availability for 2024?
I’m not referring so much to the global shutter part, but rather that with no green LEDs necessary anymore, the coprocessor and camera “in-a-box” is that much less necessary. Granted, the power regulation and cable routing for cameras becomes something to deal with, but it allows you to put your coprocessor near your other electronics, and run multiple cameras out of it.
This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.