Global Mapping Algorithm for a Driverless Race Car

In autonomous driving systems and robotics, one of the main characteristics is the capability of the system to localize itself in the environment, to create a map and to optimize the navigation, which is the ability of getting the vehicle from place to place. The Simultaneous Localization and Mapping (SLAM) is the branch of the autonomous system responsible to solve this problem. Fusing the data coming from sensors, it can provide a local and a global map, beside to correctly localize and orient the vehicle in it. The aim of this thesis work is the deployment of a robust global mapping algorithm able to provide a precise and accurate map, of an unknown environment with cones, to the control system and allow a safe and optimal navigation. The autonomous system is tested and experimentally validated both with simulations in ROS environment with Gazebo and in real time using an RC model of the SC19 electric racing vehicle developed by Squadra Corse PoliTO. The sensors used for the perception pipeline are the ZED stereo camera and the Velodyne LiDAR, while for the localization the SBG Ellipse-N inertial navigation system is adopted. The final aim of this thesis is to develop a reliable and efficient autonomous system that can be used by the SC19 vehicle to participate to the FSD (Formula Student Driverless) competition.