Evaluation of the modern visual SLAM methods

Abstract

Simultaneous Localization and Mapping (SLAM) is a challenging task in robotics. Researchers work hard on it, so several novel SLAM algorithms as well as enhancements for the known ones are published every year. We have selected recent (2013-mid. 2015) approaches that in theory can be run on mobile robot and evaluated it. This paper gives brief intuitive description of ORB-SLAM, LSD-SLAM, L-SLAM and OpenRatSLAM algorithms, then compares the algorithms theoretically (based on given description) and evaluates them with TUM RGB-D benchmark.