A Cognitive Risk Assessment Framework for Human Errors in Level 3 Automated Vehicle Takeover Events

This study develops a model to identify and assess human errors during the takeover process of Level 3 automated vehicles. Hierarchical task analysis (HTA) defines key driver tasks, while the information processing model (IPM) divides the takeover process into perception, decision, and execution stages to identify potential errors. Failure mode and effects analysis (FMEA) combined with evidence reasoning (ER) is used for risk quantification. The results identify critical errors such as being out of the loop, misinterpreting information, incorrect actions, fatigue, and delayed responses. Although the decision stage shows slightly higher risk, overall risk levels across all stages are similar, indicating strong interdependence among perception, decision, and execution processes. These findings suggest that takeover safety depends on integrated system performance rather than individual stages. Comprehensive safety strategies integrating perception assistance, decision support, and execution feedback are essential to improve the safety of Level 3 automated vehicle systems.