Design of the Safety Case of the Reinforcement Learning-Enabled Component of a Quanser Autonomous Vehicle

Safety assurance is paramount across industries where mission-critical systems operate, mitigating risks of catas-trophic failures. Safety cases play a pivotal role, particularly in safety critical systems (e.g., autonomous vehicles), in ensuring system reliability and acceptability, providing a structured argument supported by evidence. However, in the safety case literature, it is challenging to get access to a complete safety case, which is crucial for the research community to contribute in this domain. Hence, in this research, we propose an approach to create a safety case for ML-enabled autonomous vehicle, specifically, the Quanser Qcar. We present a complete safety case for a reinforcement learning algorithm applied on the Quanser Qcar to avoid collisions in an unsignalized 4-way intersection. Finally, we report the lessons learned and provide the safety case for the research community to reuse.