Hello,

Suppose I have a map of the environment modelled with an occupancy grid obtained by registering 1000 2D-LIDAR scans. Let's call this map : the global map.

Suppose now that a local map (occupancy grid) describing a certain location of the same environment is built (online) using only few 2D-LIDAR scans, say 10.

Does it make sense to try to match the local and the global maps in order to perform tasks such as localization/loop-closure ? (especially if this matching uses features such as corners)

If corners are used in the matching process, the global map would provide more "reliable" corners since this map contains much more scans (1000 vs 10 in the local map). Is there a way to have the same detected corners in both maps ?

Regards.