Path planning and navigation is one of the most important aspects of the mobile robot. In this thesis, the problem of navigation through an unexplored region is discussed. The process of navigation consists of the local navigation and the global navigation. The local navigation involves both learning the information about the robot world and collision avoidance. To take the safety of robot into account, the local navigation keeps robot preserving the given safety distance from obstacles. The global navigation accomplishes finding of a path which is apart from obstacle while maintaining the safety distance by using the vertex-map. As the navigation process repeats, the resulting path converges to the shortest path. The vertex-map that models the robot world is incremently updated by the local map of local navigation. Though a partially incorrect vertex-map is constructed by uncertainty of ultrasonic sensors, vertex-map can be improved by local navigation. To obtain the data of environment, multi ultrasonic ranging(USR) system is developed. This system consists of five ultrasonic sensors. The result of experiment show the possibility that multi USR system is applicable to any free ranging mobile with 2-dimension mobility.
