[ozrien]

Hey Jefferson, I think I have it figured out. Seems like calling joystick routines in the disabled loop has the undesired effect of causing the RIO's backend to crash/restart on power boot. It only happens once and doesn't affect much since the robot is booting up anyway. But there is a moment where when the CANTalon's object constructor's set up their messages, there is no one listening at the other end.

If you're grabbing joystick values in disabled() and using CAN Actuators I recommend calling one of the setXXX() routines on each CAN Actuator in the disabled loop. That way if this problem occurs, the code will periodically signal to the back end that those CAN objects exist and should be enabled. I reproduced your problem and tested the workaround.

So for your two CANTalons (or just do all of them), try calling set(0) in the disabled loop so that on-power-boot they get reregistered ok.