I am trying to find the most optimal camera mounting location (height and angle) for the 2024 field. From last year, the tips I learned are that the camera shouldn’t be at the same height and should be angled up or down by 10-15 degrees. With this year looking like it will have a lot of smaller robots, our team was looking at doing a low-mounted angled-up camera.
This year’s April tags are posted at similar heights:
Source: 122 cm
Speaker: 132 cm
Amp: 122 cm
Trap/Stage: 121 cm
Assuming you’re using 3d processing… you’ll definitely need to angle them up, but when it comes to 3d processing, especially with multiple tags in frame (as is the case for most important spots here), you additionally want to tilt the camera(s) on multiple axes if you want the most accurate results from the SolvePnP multi-tag algorithm. I think one team did this last year and it worked out very well.
Now, if you’re doing 2d, just do a thing pointed up like 10° and straight in front of the robot.
Yes we are going to do 3d processing to get localization. We planning on doing pitch already, should we do roll or yaw for the other axis?
Yes. I think 6328 did it too so you can check them out. Especially with the new Multi-Cam support in PhotonVision you’re likely best off going for 2 cameras rotated all about all crazy on opposite sides of the robot, with both still pointed at the tags.
4481 is currently going for 2 limelights, 1 on the front and 1 in the back of the 2024 robot. This guarantees the robot to see enough apriltags during our auto/teleop to accurately track our position on the field. We’re going for 40cm (15.7 in.) above the ground to make sure other robots don’t block our vision.
This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.