Particle Filter Based Fast Simultaneous Localization and Mapping

With respect to the necessity of more autonomous capabilities for mobile robotics in ubiquitous terrains, capable methods on localization and mapping have been developed for the last decade. One of these methods is the Fast SLAM approach which is an extension of the original SLAM problem suggested. In this report we are expressing our methodology for applying a Fast SLAM method based on particle filters applied on the generating map. Differing from its continuous space relatives, in particle filter based Fast SLAM we are considering discrete samples in the robot world covering map poses, sensor and motion models. In our experiments we have used a simulation platform consisting of 3 different map types for a mobile robot existing in a 2D world. Keywords— Fast SLAM, Particle Filters, Sensor Model