Developing a Japanese Driver Model Based on Physiological Responses and Driving Behaviors in VR Disaster Simulations

​

This research is part of a Japan-India joint initiative on disaster resilience aimed at improving road safety under extreme weather hazards like heavy rainfall and landslides. While virtual reality (VR) simulations are highly effective for safely replicating such hazardous environments, previous driving behavior studies have focused almost exclusively on visible parameters like speed and collision avoidance. To address this gap, this project integrates warning system uncertainty and cues with the driver's internal cognitive and physiological stress. Specifically, this study focuses on building a Japanese VR disaster driving scenario and capturing real-time driving behaviors alongside biometric signals (Electroencephalography [EEG] and Heart Rate Variability [HRV]) from Japanese participants using a Meta Quest 3 headset, Logitech G29 steering controls, and advanced physiological sensors. By modeling the physiological and behavioral responses of Japanese drivers, the research ultimately aims to construct a comprehensive driver model and develop a stress-adaptive reinforcement learning AI agent. Through these phases, the project contributes to the understanding of cross-cultural human responses to disaster warnings and supports the development of adaptive driver-assistance systems for enhanced crisis management.