Comparative analysis of ROS based 2D and 3D SLAM algorithms for Autonomous Ground Vehicles

Abstract

Simultaneous localization and mapping algorithm (SLAM) is a technique for estimating sensor motion and reconstructing the structure in an unknown environment. SLAM is extensively used in the concept of autonomous driving or navigation which helps robotic vehicles to move autonomously. This paper presents a comparative analysis of different ROS based 2D SLAM algorithms such as GMapping, Hector SLAM, Karto SLAM and 3D SLAM algorithm such as Real-Time Appearance-Based Mapping. Scenarios with a scaling factor resembling an underground parking was built in real-time and simulation for analysis and validation purposes. The maps generated by applying the respective SLAM algorithms in real-time and simulation was analyzed and validated quantitatively based on quality and accuracy using the Structure Similarity Index.