Abstract
Robust self-localization in a dynamic environment is an important ability for autonomous mobile robots to accomplish the tasks efficiently. There are several researches to realize robust self-localization though most of the previous methods were unambiguous or had large computational cost to estimate the position. In this paper, a self-localization method that realizes robust self-localization with limited computational resources is proposed. Proposed method switches the estimation reactively considering number of detected landmarks and estimation accuracy. We verify the practical effectiveness of proposed method applying to the RoboCup Middle-Size League.
