Fuzzy Map Matching for the Atlas SLAM of Mobile Robots

A novel approach of reducing computation complexity for simultaneous localization and mapping (SLAM) of mobile robots, which based on self-organizing fuzzy neural networks (SONN) was proposed in this paper. The matching component for local maps in the atlas SLAM system is replaced by an SONN. Our approach is superior to the original one by taking all of the feature parameters in a local map instead of a set of map signature elements. In addition, the computation cost which is required to determine the active local map and to close loop was reduced to a reasonable size.