Revision history [back]

The standard way to do achieve this is by using tf. Approaches that are able to localize the robot like gmapping or AMCL publish the map -> odom transform. If you follow this convention with your localization approach, you should be able to use it as a drop-in replacement for the other mentioned approaches. I did that myself with hector_mapping and it works without problems.

In fact, move_base doesn't care at all which localization approach is used, as it only uses tf to do queries about transforms. See REP 105 for an overview of coordinate frames for mobile robots. Again, the job of your localization node would be providing the map -> odom transform.