Perception and localization often depend on each other. When building a map, a model of the environment is being constructed and perception and localization cooperate in order to put pieces of the map in the right place relative to each other. When using a map to localize, the robot is determining its location based on matching what it sees to what it expects to see and, in this case, localization depends on perception. More sophisticated approaches can construct a map and localize from it at the same time.

This section will present many aspects of both making and using maps of the environment. Many mobile robots need some kind of map in order to function effectively. Maps are needed when accurate absolute pose is needed. Accurate pose is needed when the robot is told to go to the coordinates of a specific place or when the robot wants to use information of any kind that is located by its position in a second map.

If localization is imperfect, it may be advisable to register data generated at different times and places. This is a more fundamental operation than either localization or mapping. It achieves both the goal of making the map consistent and the goal of refining the estimate of the intervening motion. For this reason, all three topics of mapping, registration, and ego-motion are presented together here.