Risk Assessment Method for Autonomous Vehicles Violating Safety Common Sense Based on Driving Behavior

As autonomous driving technology continues to evolve, ensuring decision-making safety in complex traffic environments has become a major challenge. This study proposes a novel risk assessment method based on driving behaviors that violate safety common sense. Driving behaviors that violate safety common sense are defined and categorized, and a comprehensive risk assessment framework is established. By analyzing real-time trajectories and driving behaviors, risk indicators are quantified to form the basis of the risk assessment model. Combined with environment safety entropy, the comprehensive risk assessment is obtained using the subjective Analytic Hierarchy Process and the objective entropy weight method to enhance assessment accuracy. The comprehensive risk levels are classified using a Gaussian Mixture Model (GMM) optimized by K-means. Finally, simulations and real-world road tests validate the effectiveness of the model. This method addresses the limitations of existing approaches by providing a more adaptive and accurate framework for assessing unconventional yet risky driving behaviors. It offers new perspectives and strategies for improving the safety of autonomous driving systems and provides valuable references for future regulations and standards.