Hello, I am using gmapping to build laser scan map here, while I found out that the communication between /slam_gmapping and /tf is bilateral:

Test gmapping on the bagfile, I extract the baselink position of tf-tree relative to map, and end up with the trajectory comparison as below. In gmapping case, the position of baselink starts getting more and more random after the first few seconds.

therefore it could be expected that the final map is totally messed up