A SLAM research framework for ROS

摘要

Most of the methods that try to solve the SLAM problem use a prediction-correction approach to estimate current robot pose and map and can be spitted into several groups depending on a set of sensors they use and details of their implementation (e.g. a type of map, feature usage, etc.). Methods that belong to the same group usually differ only by cost functions and ad-hoc optimizations. To the best of our knowledge, a framework that provides a common set of components in order to speed up SLAM research is not publicly available (frameworks and toolkits that simplify development of particular SLAM parts are not taken into account). The paper introduces a framework that is under development and provides a set of components that simplify creation of methods based on 2D laser scan processing. The description of tinySLAM and GMapping implementations atop of the framework is provided in order to justify its usability.