| This thesis describes a method for mapping unknown large scale static environments using a hybrid of topological and metric representations. A global map is formed from a set of local maps for sub-regions of the environments. Each local map contains quantitative environment information used to define a local reference frame. These maps are referred to as islands of reliability because they are associated with the sub-regions whose local structure is best matched to the sensors we are using. The connectivity of these islands is represented topologically. The key mapping problem we consider is where to place the islands of reliability and to what extent they should cover the environment. This is accomplished by defining the placement criteria in terms of the task to be satisfied and the uncertainties of the mapping agent. Islands are distributed about the environment at areas suitable for extracting metric information relevant to a localization task. |