Safe Reconfiguration of Autonomous Driving Systems

Autonomous driving technologies have gone through rapid development over the past few years. With the emergence of various automation modes, examining the safety of self-driving systems during mode transition is now essential to ensure the overall safety of the vehicle. The goal of this research project is to develop a novel algorithm for safely reconfiguring autonomous driving systems. Our algorithm considers critical environmental factors and provides safety guarantees as vehicles undergo the reconfiguration process. Using CARLA, an open-source driving simulator, we implement and test our algorithm by running it through an extensive set of auto-generated adversarial driving scenarios. The results indicate a reduced number of collisions and improved road safety, which verify the effectiveness of our prototype algorithm.