A COMPLETE END-TO-END SIMULATION FLOW FOR AUTONOMOUS DRIVING FRAMEWORKS

The research on automotive industry in the last year focused on three main aspects: electric engine, advanced infotainment and autonomous driving. Nowadays autonomous driving represents one of the hottest topics in the automotive field, since technology advancement in these last years, especially with the new discovery on Artificial Intelligence, allowed to lay the foundations for the development of fully autonomous vehicles and revolutionize the concept of autonomous mobility. Since building a real vehicle equipped with the needed sensors is very expensive, it is critical to test every component, both hardware and software, considering all the possible scenarios where the vehicle may operate. It is clear how Software and Hardware in the Loop play a key role for the development, allowing a preliminary validation of the vehicle before road testing. However, this approach, to be reliable, requires a virtual environment that mimics the conditions of the real world. The goal of this thesis is to create an automated system for the generation of three-dimensional maps starting from simple 2D maps. The building of a structured simulation ready map, a detailed map containing information of the environment, starting from OpenStreetMap data is the key for reduce costs and time for fast prototyping of this new generation of algorithms. The map generation is accomplished by using Carla simulator, an Unreal Engine-based simulator. After this, the map is converted into a specific format suitable for an eventual field test by using a bridge between Carla and the ROS-based platform Autoware. The last part is dedicated to the evaluation of the achieved result considering possible limitations and improvements.