Enhancing Continuous Risk Assessment: The Role of Safety Engineers in Early Hazard Identification

As Automated Driving Systems (ADS) revolutionize the intelligent transportation landscape, ensuring unparalleled safety is increasingly essential. Traditional risk assessment methodologies, primarily designed for human-driven vehicles, struggle to adapt to the complex, ever-changing environment of ADS. This paper introduces a cyclic process aimed at augmenting continuous risk assessment for ADS, addressing the limitations of existing standards, which focuses on the functional safety of road vehicles but assumes static risk and the presence of a human driver for control and responsible for safety. Our proposed process transcends these limitations by integrating learning-based risk assessment, aiding safety engineers in early hazard detection for ADS development. The cornerstone of this approach is the Plan-Do-Train-Adjust-Assess cyclic process, which facilitates continuous improvement in risk assessment under diverse driving conditions. This method leverages advanced learning algorithms and integrates risk-specific contextual information, thus bridging traditional gaps in risk assessment. Critically, the process allows for the evaluation of severity and controllability to vary across different dynamic environment. This variability is determined by factors such as the operational domain, system complexity, and the evolving risk knowledge obtained through an iterative process. The insights gained from assessing severity and controllability aid in creating and refining essential safety mechanisms.