Exploration of ruled-based and probabilistic localization algorithms for wall-following robot

Current localization algorithms for mobile robots require high computation cost and are impractical to implement in a micro controller. In this paper we propose three localization algorithms for hexapod wall-following robot with low computation power and limited sensors. Our algorithms are based on non-fuzzy rule-based classification algorithm, fuzzy rule-based classification algorithm and particle filter algorithm. We use a standard field for international robotic contest for comparing effectiveness of the algorithms. Experimental results shows that the particle filter based algorithm can classify more labels compared to both non-fuzzy and fuzzy rule-based algorithms.