Robotic perception for autonomous navigation

This thesis presents the research work the author carried on during his PhD on the topic of robotic perception for autonomous navigation. In particular, the efforts focus on the Self-Localization, Scene Understanding and Object Detection and Tracking problems, proposing for each of these three topics one or more approaches that present an improvement over the state-of-the-art. In some cases the proposed approaches mutually exploit the generated information to improve the quality of the final results. All the presented methods are experimentally validated and compared to state-of-the-art methods' results. Finally, a real autonomous car implementation is presented, that served as a test bed for the approaches presented in this thesis work.