This is my first season as Mentor for team 2478. I would appreciate any feedback/help how to best implement apriltag and object detection to our robot using an OrangePi 5 Plus(or RPi4/5) and an Arducam IMX586.
What software stack should we use? The Ubuntu versions available for OrangePi seems to be very limited compare to Raspberry Pi. Could some of the teams that uses Orange Pi share what OS and framework they are using. Or even share a github repository with code snippet to get us going?
Can OrangePi(RaspberryPi) only communicate with the RoboRio 2.0 with Tcp or UDP packages? I saw that a team tried to develop ROS2 for the robot(Team 2383), did that include pub/sub communication between RoboRio,Orangepi, (Ardu)Cam, Motors controllers? What are the main challenges of using ROS2 for FRC 2025 season?
Any feedback on some of my questions are appreciated!
Thanks
JadeDragon
The best software for vision in frc without doing a lot of custom work is photonvision. It supports april tags, object detection etc. It works on the orange pi and supports arducams
The Zebracorns, team 900, are well known for their use of ROS (seven seasons now) and have a lot of good info on their team website. The regularly publish white papers on technical topics, including ROS and have all their code public. Take a look at their Zebracorn Labs page to see what they have. I can tell you that they were using both Apriltag and object detection in 2024, so they might well have the solution you need, or at least good advice on where to start.
Yes, coprocessors have access to TCP/UDP over Ethernet. The built-in and fully integrated pub/sub infrastructure for FRC is called NetworkTables (NT), part of the WPILib project. The implementation is portable to and available for coprocessors, and the existing COTS vision solutions including Limelight and PhotonVision use NT to talk to the robot code. A few teams have successfully used ROS, but it is not a common solution for FRC and takes a lot of work to set up.