Revision history [back]

Normally, you have all nodes that you need in a system running all the time, maybe idling most of it. If some actions need to be performed, they are then triggered by some high-level executive, e.g. a python script, that sends a command to the corresponding node that starts and maybe stops processing. For instance, if you have a face-tracking node and a color tracking node, both would provide an action. The executive node would send a goal to the face tracker and preempt it as soon as it is supposed to stop tracking. The color tracker would behave similarly.

Also, a multiplexer could work that allows to enable/disable messages on a node's input topic.