During some of our matches in the ISR#3 event we’ve suddenly lost control over the robot (most noticeable in the swerve that goes crazy).
We havn’t noticed this issue while practicing at out workshop, only in the event, while connected to the FMS.
We’ve tried switching RoboRIOs and NavXs but it seems that it didn’t solve the problem.
We thought that it was related to a known problem with the I2C port on the roboRIO that causes the RoboRIO to stop responding when a device is connected through it (like our colour sensor), but after talking to other teams who had experienced the issue with the I2C it seems that we think that this isn’t the source of our problem.
I’ve attached the log files from two different games, one from a game where the issue did not occure (15_45_13, normal), and one from a game where the issue happened (15_15_22, abnormal/happened). logs.zip (283.3 KB) Capture|675x500
The thing that pops out from the log image (sorry, I don’t have a DS here to open the log files) is very high CAN utilization. It might be a that your CAN bus is overloaded and some messages get lost. though, I am not sure why you are not encountering it more often.
I2C - You are right to assume it’s not that as the RIO would stop responding and the robot will stop until a reboot is done.
Actually - we had I2C issue, and the results were from the robot stopped functioning altogether to stopping responding to specific commands. In some cases the connection was still ON in driving station. So, I would not be dismissive of it.
The issue becomes MUCH worse if you query the I2C port when nothing is connected to it.
Anyway - adding NavX immediately fixed it. No disconnects or random behavior.
CAN overload is certainly possible as you need quite a few motors for swerve. But from our experience unless you consistently overrun the loop, we did not see any ill effects keeping CAN to 90% all the time, with occasional peaks (we do not run swerve, but we have a good number of CAN devices).
It looks like about halfway through the match the roboRIO CPU spikes, it stops reporting it’s enabled state as frequently, and then a bit later it quits altogether. I have to imagine all those are related, but without being able to read the errors and see the robot in person it’s hard to say. I’ll be one of the CSAs at DCMP next week. If you ask us to take a look then, I’d be happy to help
This might be a good thing to ask the inspectors, they often have experience with this sort of thing.
Worst case scenario if your bot goes fully rouge, the judges have a kill switch
