the question of merging maps

[liangyuan109]

hallo,i have a question,I've recently been testing map merging.what algorithm is used for merging maps? Why do two maps sometimes merge successfully and sometimes fail to merge during testing?
looking forward to your answer!

[matlabbe]

You can check this paper for more details: https://arxiv.org/abs/2403.06341

Map merging is vision based. When a loop closure is detected between two sessions, we add a new constraint linking the two sub maps. During graph optimization, they will optimize together.

If two maps don't merge, it would be mainly a lack of visual features to estimate the transformation of the loop closure. Ideally, to improve loop closure acceptance, make sure the robot/camera is revisiting the same area with very similar point of view, to maximize the good visual correspondences.