Virtual vehicle sensor's set-up and synchronous logging from CARLA Simulator

Improving road safety is a global priority, with the World Health Organization identifying human error as the main cause of accidents. In response to this, the Zero Vision project has been launched with the ambition of completely eliminating road deaths by the year 2050 through the mandatory introduction of Advanced Driver Assistance Systems (ADAS) technologies and the development of autonomous driving systems. Before being deployed in vehicles, these systems must undergo rigid testing and validation phases to ensure their effectiveness and safety among the variety of conditions under which they may operate. This process requires the collection of huge amounts of data, even for complex driving scenarios. As such collection is costly and risky when dealing with borderline situations, the work proposed in this thesis aims to develop a fully customizable virtual environment for the generation of synthetic data which can be used to integrate real-world data in the validation phase, enabling the safe and controlled reproduction of the most difficult scenarios to be captured on the road. The platform allows complete customization of every aspect of the simulation, including weather, traffic conditions and detailed configuration of each virtual sensor installed on the simulated vehicle. In addition, the data collected will be organized according to one of the most widely used automotive standards, ensuring compatibility with validation and development processes already adopted in the industry. Although synthetic data can not completely replace real data collected on the road, through this work, an effective methodology is proposed to accelerate the validation of ADAS systems, contributing to the progress towards autonomous driving.