Review of Autonomous Driving in Unexpected Events

In the field of autonomous driving, much of the research on decision-making systems focuses on predictable data. This paper emphasizes unpredictable data and behaviors, categorizing them into three types: (1) erratic behavior caused by issues with the driver, (2) behavior resulting from vehicle malfunctions, and (3) unpredictable driving behavior due to road conditions. Based on these categories, this paper establishes a database encompassing these three types of behaviors, which can be used to provide data for testing related algorithms. The analysis and research in this paper demonstrate that the ability of autonomous driving decision systems to assess unpredictable behavior is crucial for enhancing driver confidence and reducing traffic accidents.