Revision history [back]

To summarise the discussion in the form of an answer. The OP is using aruco markers to determine the transformation between the camera frame and the world frame, however the markers were occasionally being occluded which meant the transformations were not being publishing and the TF tree would be occasionally broken. Importantly in this case the camera is fixed so the TF can be assumed to be static.

A potential solution would be making a simple custom node which averaged a few seconds of the world to camera TF at the start of the experiment and stored it. The node can then publish this stored TF at a fixed rate and it will not be effected by any occlusions of the marker.