Abstract Next generation planetary rovers will require greater autonomous navigation capabilities. Such requirements imply the management of potentially large and rich georeferenced data sets stored in the form of maps. This paper presents the design of a data management system that can be used in the implementation of autonomous navigation schemes for planetary rovers. It also outlines an approach that dynamically manages a variety of data content and the uncertainty of the spatial relationship between two maps; in addition the proposed framework provides basic path planning operations through maps, and the correlation of maps in localization operations. Timing results from a rich data set demonstrate the efficiency of the proposed framework. In addition, experimental results on the usage of our Atlas management system by a rover performing autonomous navigation operations are also presented.