Obstacles Detection and Global Mapping Algorithm for an Autonomous Racecar

Nowadays, mapping is one of the most active research fields in robotics. In this thesis it is faced the global mapping issue related to the SLAM (Simultaneous Localization and Mapping) process of an autonomous vehicle which will participate to the FSD (Formula Student Driverless) competition. The mapping process together with the localization one allow to perform car navigation, which is the ability of getting the vehicle from place to place. Thus, according to Leonard and Durrant-Whyte H. “Dynamic map building for an autonomous mobile robot” paper, SLAM allows to answer to three fundamental questions for a mobile robot: “Where am I?”, “Where am I going?” and “How do I get there?”. To do this, it was decided to adapt to our case the algorithm contained in the PerceptionAndSlam_KTHFSD1718 package of the KTH Royal Institute of Technology FSD team published on GitHub website. The thesis can be divided into five sections. The first one deals with the state of the art of the SLAM process, the second one deals with the sensors setup with which the car is equipped, the third section faces the localization and the visualization perception processes and how they are implemented by our team, the fourth section speaks about the implementation of the mapping algorithm using Python 3.7 as programming language and ROS Melodic (Robot Operating System) installed on an Nvidia Jetson Xavier computer mounted on the PoliTo SC19 electric racecar. Finally, in the fifth and last section, it is faced the testing of the mapping algorithm through the EUFS simulator software and some experimental sessions on the Torino Aero Club racetrack.