Performance comparison of 2D SLAM techniques available in ROS using a differential drive robot

Abstract

This document presents a performance comparison of three 2D SLAM techniques available in ROS: Gmapping, Hec-torSLAM and CRSM SLAM. These algorithms were evaluated using a Roomba 645 robotic platform with differential drive and a RGB-D Kinect sensor as an emulator of a scanner lasser. All tests were realized in static indoor environments. To improve the quality of the maps, some rosbag files were generated and used to build the maps in an off-line way.