Robust mobile robot on-the-fly localization: using geometrical feature matching method

Self-localization is an essential component of any autonomous mobile robot system. This paper presents a fast and robust feature-based approach to mobile robot on-the-fly global localization in dynamic environments. A robust feature extraction and correspondence algorithm was introduced to extract the landmarks of the environment and to triangulate the robot's positions. The approach take into account the hard real time constraint and the partial observability of the environment. On-line localization is performed by matching candidate landmarks from the robot's local view to the tracked landmarks. Experimental validation in the context of middle size league RoboCup was carried out in a real environment with the mobile robot Jiaolong which equips a laser range finder (LRF).