Closed aedsinger closed 11 months ago
@aedsinger Does this change the behavior of nodes (e.g. stretch_demos) that interact with the action server?
It should not. We've verified through this gist that the pimu trigger motor sync issue is resolved: https://gist.github.com/hello-fazil/5c0732e33eb140fc5a717f339218d100
This moves the push_command call of Robot to single place - the main loop of stretch_driver. Commands can come from callbacks (eg runstop), base velocity control, or the Joint Trajectory Action server. They now depend on the main loop (executing at 15hz) to actually push the command.
The reason for this is to avoid push_commands that occur close together in time, which causes an issue with Pimu motor_sync. The downside is that a joint command may wait a bit longer for the command to actually get pushed to the robot (eg, 1/15hz max).