Revision history [back]

You would need two different tf frames map->kinect1 and map->kinect2, which you can publish statically. The static transform publishers can run all the time. Then, configure two launch files for the kinect cameras. One should publish in the kinect1 frame, the other in the kinect2 frame. If they both output to the same topic but with different tf's, octomap_server will lookup the correct tf to the global /map frame.

You can also publish kinect1 directly in the map frame, and publish a static transform map->kinect2 (= kinect1->kinect2).

You would need two different tf frames map->kinect1 and map->kinect2, which you can publish statically. The static transform publishers can run all the time. Then, configure two launch files for the kinect cameras. One should publish in the kinect1 frame, the other in the kinect2 frame. If they both output to the same topic but with different tf's, octomap_server will lookup the correct tf to the global /map frame.

You can also publish kinect1 directly in the map frame, and publish a static transform map->kinect2 (= kinect1->kinect2).kinect1->kinect2). But you definitely need the two kinect frames (whether static or moving while in RGBDSLAM) registered in the global map frame.