A Robust Framework for Localization in Autonomous Systems

Abstract

In recent years, there has been a growing interest among researchers in newlineAutonomous Mobile Robots. In order to move autonomously, a fully automated robot must newlinebe able to locate itself in its moving environment. The process of determining the position newlineof an autonomous system within specific location or surroundings is termed localization. It newlineis essential for generating accurate maps and facilitating efficient path planning for newlineautonomous systems. This process is an essential step in autonomous system navigation, newlineallowing the system to make decisions based on its current location. During the localization newlineprocess, the robot uses data gathered from its deployed sensors to estimate its position. The newlineprimary aim of localization is to minimize ambiguity in location information, enabling the newlinerobot to navigate its surroundings efficiently. newlineThe most commonly used technology for enabling localization in outdoor newlineenvironments is the global positioning system. However, the accuracy of position estimation newlinereduces significantly in GPS-denied environments, especially in urban areas. Other newlinetechnologies contributing to localization in autonomous systems include Light Detection and newlineRanging (LiDAR), which identifies obstacles by illuminating targets with laser light, and newlinevisual odometry, which utilizes data from onboard cameras. Creating robust and accurate newlinelocalization methods is an important area of research in autonomous systems. This thesis newlineaims to reduce uncertainty in the localization of autonomous systems. We conducted newlineexperiments on the KITTI benchmark data to demonstrate the efficiency and improvement newlineof the proposed algorithm over previous work. The dataset was compiled by performing newlinedriving tests in different traffic scenarios in Karlsruhe, Germany newline