Frank Bahrmann, Sven Hellbach, and Hans-Joachim Böhme
Autonomous Robots, Fuzzy Sets and Systems, Indoor Navigation, Localization
This paper presents a novel approach to cope with a major problem in long term mobile robot operation - an adaptive localization. We achieved this by utilizing a Takagi-Sugeno Fuzzy Inference System to generate a model of the environment which learns the world's dynamic behavior. Due to the fast inference system it is possible to keep this model up to date in real time with very low computational effort. We show in simulation runs that the model is stable in different degrees of dynamic environments over hundreds of kilometers. Furthermore, we demonstrate its application under real world conditions. To improve traceability we provide data sets of different experiments.