The implementation of the design of any navigation system depends on available hardware and the intended application. A robotic navigation system can be split into four problems: Localization, mapping, obstacle avoidance, and a controller. In this dissertation, the current state-of-the-art for research in localization and obstacle avoidance is presented. Also, the Argo class rovers and the MarsDome Experimental Facility are described in detail, including the modular philosophy of the Argo software architecture. Then, the design and implementation of the Argo navigation system is provided, with greater emphasis on the visual localization, odometer localization, and obstacle avoidance modules. The results of several experiments demonstrate that each module is efficient and is sufficiently robust to be successful in the integrated navigation system. This research is also put in context by explaining some of the related work being conducted at the Space Robotics Group and its relevance to collective mobile robotics research.