Mobile robot navigation using hybrid simplified map with relationships between places and grid maps

Uncertainty-based localization in a topological robot navigation system

“…The following works, that belong to the third group, directly built the global topological map and the local metric maps without the need of a global metric map [21,46,57,63,90,100,120,127,128,156,159,173]. Estrada et al presented Hierarchical SLAM in [57].…”

“…In [159], each node of a topological map contained the set of images in that region and it was augmented with metric information at places (nodes) where the robot turned. Similarly, in [156,46,173], metric information was not stored for the whole environment. In [156], Tomatis et al presented a localization and map building method based on a topological global map and they added local metric maps only for some specific areas of the environment.…”

“…In [46], Dayoub et al presented a method in which each of the nodes of a pose graph was associated to a 3D point cloud and the point clouds of interesting areas such as intersections could be merged together in a 3D occupancy grid. In [173], metric maps were only created for complex areas, whereas corridor and simple areas were maintained through a topological map. Nitsche et al [120] proposed a hybrid mapping method consisting in a topological map and fixed-size metric submaps.…”