Advanced Methodologies for Safe and Secure Software-Driven Autonomous Vehicles

The automotive industry is rapidly evolving due to advancements in sensors, wireless networks, and Artificial Intelligence. Modern vehicles are increasingly connected and autonomous, with systems that require extensive connectivity to interact with infrastructure, other vehicles, and cloud networks. Moreover, software is assuming a central role in managing these technologies. This shift creates new Cyber-Physical Systems (CPS) that have the potential to reshape transportation, but also increases complexity, making cybersecurity, regulatory compliance, and safety critical. This work explores key challenges in vehicle safety, offering contributions across multiple layers of the automotive software stack. At the lifecycle level, it presents methodologies for integrating safety and security into processes. At the application level, it introduces an approach for autonomous driving in critical scenarios and a method for verifying neural network robustness. Additionally, emerging threats in the in-vehicle network, especially the CAN bus, are examined. The results from the conducted experiments highlight promising directions and underscore some of the most pressing issues to address in the near future, paving the way for further research in the field.