Where to Look? VR Gaze-Based Analysis in Adverse Weather Driving

Driving under visibility-impaired conditions significantly increases the occurrence of accidents. To understand why accidents are more frequent in adverse weather, it is essential to examine how drivers redistribute their visual attention during emergency situations under different weather conditions, and how this redistribution differs depending on whether an accident occurs. To address this issue, we developed VR driving simulations that incorporate varying weather conditions and accident types integrated with eye-tracking. We analyzed gaze allocation from 50 participants across a 10-second window preceding a critical event, segmented into 1-second bins. Comparing successful and failed trials under both clear and adverse weather conditions, we identified significant differences in the timing of attention re-allocation based on task success and weather. Drivers who failed to complete the scenario fixated more on the forward view during the early and middle phases, with differences between the Pass and Fail groups emerging earlier under adverse weather conditions. Our findings highlight early imbalances in gaze behavior as potential indicators of successful hazard avoidance, which could contribute to the design of gaze-based driver assistance systems aimed at preventing accidents.