Abstract
Autonomous mobile robots has been an active research recently. Especially, laser range finder (LRF) based navigation has rapidly improved. On the other hand, in spite of intensive studies on vision based navigation using cameras, they are not successful in real world. In this paper, we develop a vision based navigation method using topological approach to reduce prior information and enhance scalability of the map. Topological map has advantage in scalability because it contains only information about impressive places (i.e. intersection) and their linkage with roads. The robot is navigated using two primal techniques: vision based road following and impressive place recognition.
