I have a rpi4 running a wpilibpi image with a model from axon on it. The detection is working fine, and when I go to the wpilibpi.local/1182 port, the detection shows up (it detects the ball, when I move the video stream tracks the ball, etc.). However, sometimes when I start up the robot or deploy new code, the ML/detections networktable is not generated. I’ve checked this with outlineviewer, and the table simply does not generate. Anyone else having this problem/know how to fix?
Most likely it’s a networking issue causing NT to not reconnect to the robot code for some reason. How is your Pi Ethernet connected into the robot network? There can be issues using the second port of the radio; I would recommend using a switch if you aren’t already. Also coprocessors tend to work a lot better with static IP configuration.
I actually figured it out just 10 minutes ago, all I had to do was connect to the pi, then go to the vision tab where I can see the script running, then put it down and bring it back up. Don’t know why I didn’t try before but it made it reconnect to network tables, so it works.
My pi is connected via Ethernet to the radio. And about the coprocessor, I’ll make sure the pi is on a static config. Thank you!