1. Introduction
Visual systems for mapping and localising in an environment are a key technology for many real-world applications, such as automatic driving service robotics, augmented reality and object modeling. There are two general purposes that map-making serves: accurate localization within a local area for tasks such as obstacle avoidance or object manipulation; and global connection information for longer-range navigational planning. One proposal for how maps could work in both cases is to consider them as manifolds; that is, locally Euclidean but globally topological [4].