Emergency Takeover under Low Driving Experience: A Study on Driver Performance Analysis and Adaptive Behaviors

Emergency takeover is a common form of human-machine interaction in existing advanced driver assistance systems. It is significant to Study the behavioral characteristics of drivers during emergency takeover events. However, few studies have considered the differences in driver's demand for Time-to-Request Lead Time (TORlt) under different risk element compositions and the influence of experience accumulation on interaction requirements. In this study, a 3×3 lane-changing obstacle avoidance simulation experiment was designed with TORlt and vehicle speed as variables. A total of 18 drivers with low driving experience participated in two rounds of experiments with the data recording. We employed three dimensions of takeover time, quality, and experience. This study selected eight indicators, including perception and decision time, exposure ratio in risk areas, and minimum collision time, to assess drivers' performance during emergency takeovers. The analysis examined the differential impact of various experimental variables, experience accumulation on these indicators, and drivers' compensatory behaviors across different scenario designs. The findings reveal that the interaction between different vehicle speeds and TORlt designs results in dynamic driver takeover performance fluctuations. Although the accumulation of takeover experience enhances drivers' performance, it also elicits negative adaptive behaviors to some extent.