Restarting ros2_control must be followed by a restart of the RAPID program

[tgaspar]

I observed an issue with the driver/communication between the ROS side and the RAPID side when the connection is lost. Most notably, if the ros2_control is restarted while the RAPID program is already running will make it so that the robot no longer moves when the commands are sent.

To reproduce (with RobotStudio):

• Start the RAPID program that has EGM communication/control
• Start the ros2_control nodes
• Trigger a motion and wait that it concludes
• Restart the ros2_control nodes
• Trigger the motion again --> the robot does not move

reconnect.mp4

To get back into a controllable state, one needs to restart the RAPID program (stop, pp to main, start).

It is important to mention that I am using slightly modified lauchfiles. But the underlying code is the same. At this point, I did not have the opportunity to dig deeper why this happens. I also didn't try the exact same steps on the real robot but I do recall having some issues with restarts.

Did anyone note similar behavior?