When the Limelight camera on our robot gets knocked out of position during a match, it disrupts our alignment and tracking. What methods or tools do you use to quickly and accurately recenter your Limelight? Are there any preventive measures or adjustments to secure it better during gameplay?
2 Likes
How are you mounting it? 2928 typically uses rigid mounts that have no adjustment once we decide how we want the camera to be positioned. Last year, the mounts were 3d printed and multiple identical copies were made. During an off-season event, the mount was broken, and we were able to reinstall it with a new mount with no issues.
7 Likes
We are currently using an angle-adjustable holder for the Limelight camera. The adjustment mechanism consists of a metal bolt and nut, which allows precise angle positioning. However, the rest of the holder is made of 3D-printed plastic, making it less durable. During high-impact situations or frequent adjustments, the plastic components tend to break easily, causing the system to fail despite the sturdiness of the metal parts.
Once you figure out the angle you want, could you use a digital protractor to find that (currently adjusted) angle and make a fixed-angle piece of solid-ish plastic to mount it?
6 Likes
Once you have an angle decided, do you still need the adjustment? Could you design a mount that matches the desired angle without adjustment? That should make it more sturdy. For example, use the existing mount to find what angle works best, measure the actual angle, then design a stationary one that matches the position.
What materials are you printing the parts out of? I have had success with PLA parts, but I’ve been moving in the direction of PA-CF. 3D printed parts should be able to survive bumper on bumper collisions, but most 3d printed mounts will break if they are the contact point.
2 Likes
Design a stiffer mount. For as much as stiff mechanisms matter, stiff camera mounts matter more. The limelight mount should be the stiffest thing on the robot, its accuracy is more important than the mechanism it is sending data to, because slight misalignments in the limelight position can cause large errors.
2 Likes
I actually think rigid mounts may be worse than compliant mounts.
We used onyx for our camera mounts, but i’d guess over time collisions could deform the mounting structure, making every future reading at a competition incorrect.
But yeah, i’d find a good angle (maybe with prototype), and then print a final rigid, non-configurable part .
My understanding is that 6328’s camera mounts are compliant because they’re exposed on their swerve pods, and they can deflect and bounce back after an impact, rather than snapping off. For a camera mounted higher / inside of the robot, a rigid mount is better.
3 Likes
Bro I can’t even recenter myself when I’ve fallen off.
12 Likes
I don’t even understand the question (though maybe because we don’t use Limelights). Isn’t the right question “How do I make mounts that don’t fall off?”
I agree with the comment that once you figure out the ideal positioning of the camera, 3D-print something that isn’t adjustable (if that was the problem). We made our initial camera mounts in 2024 with several holes so that we could find something that worked well, then later versions didn’t have any alternate choices at all.
One tip (that we didn’t do last year, but should have): if you have mirror-image mounts on either side of the robot, put the “L” or “R” or something other indication of where they go right into the CAD.
1 Like
Also, add a date or version number or angle or similar info that you might want to reference quickly if you’re thinking you might change it later. That way you can tell which of the 20 camera mounts floating around is the version you’re looking for without needing to measure.
1 Like
My general rule of thumb is: “Anything that CAN adjust WILL adjust, usually when you least want it to.”
As others have said, find the angle you want, then 3D print a hard mount.
For that matter, usually I don’t even deal with the adjustment on a physical camera, we mock up the camera location in CAD with an estimated FOV model, then use that to get the rough position we’re looking for and the programmers figure the rest out. If you’re robot is already in CAD, it’s pretty easy to download the field model and drop it in to estimate positions.
I suppose if you were trying to get a really specific angle to avoid an obstructing component on the robot I might consider a temporary adjustable mount, but usually we just print a whole new model if something is off a bit.
1 Like
You just reminded me of an incident from a few years ago.
If you look at 1197’s performance In Galileo 2016, we were inconsistent to say the least with our shooting, 4-6 record in quals. But, what you might notice is that we suddenly improved late on Day 2. What happened?
Our camera was on an adjustable mount. Somehow, that mount had shifted between our previous event(s) and Galileo quals. This meant that most of our shots were off-line and if you know anything about our shooter that year it was a LINE shot, and hard. We’d been troubleshooting for a while when this was noticed; after realigning and redialing, we started hitting more shots. Put together just enough to be noticed by the #1 alliance (and I think borrowing 987’s “target on a pole” helped out with that), and powered our way to Einstein.
We’ve used “That ain’t goin’ NOWHERE!” mounts for the camera ever since.
1 Like
Even non-adjustable camera mounts are problematic, in 2022 division elims, my team suffered a sudden accuracy decrease due to a poor limelight mount. Our shooter was designed to be incredibly stiff, but we used a limelight on a pole to avoid camera blocking defense and to extend our FOV, which lead to stiffness issues but not enough care was put into it. Some combination of the tube yielding, screws slipping (because screws are not pins) and the 3d printed CFny mount deforming caused our auto-aim to fail and miss shots, forcing us to play sole defense instead of a hybrid strategy, contributing to a pair of losses an an early elimination.
I would generally suggest a solid metal or composite camera mount which is strong and stiff enough to not yield under impact or deform significantly under acceleration, and put significant care into designing in proper locating features to reduce any tolerance stackup that ends up reaching the camera. Fractions of a degree matter, especially with the simpler tracking algorithms.