Analysis of Computational Need of 2D-SLAM Algorithms for Unmanned Ground Vehicle

Abstract

The need for advancement in the robotic car or driverless car or autonomous vehicle is vital, as safety concerns and advancements in automotive technologies raise have led to market penetration and acceptance of driverless cars. Simultaneous Localization and Mapping (SLAM) is one of the important features of autonomous transportation. SLAM helps us solve the problem of a vehicle maneuver in unknown locations. It assists the vehicle to understand the surroundings (mapping) and its current position in it (localization). SLAM algorithms can be implemented in autonomous vehicle application using the Robot Operating System (ROS). In this paper, four of the commonly used SLAM algorithms, i.e., Gmapping, Hector SLAM, Karto SLAM, and RTAB are implemented on an autonomous vehicle for computational efficiency. Comparison study of computational resource utilization of all these algorithms is done in both simulations and Real-time implementation.