Toward Personalized Fitness-to-Drive Assessment Using Driver Digital Twins

Abstract

Enhancing road safety while maintaining a comfortable driving experience remains a paramount challenge for the automotive industry. This research addresses the lack of personalized yet effective real-time monitoring and alert systems by developing a driver-specific Fitness-to-Drive (FtD) index through the implementation of a Driver Digital Twin (DrDT), built as a specialization of a Human Digital Twin (HDT). The proposed FtD algorithm integrates critical driver-specific factors such as age, driving frequency, experience, visual and cognitive distractions, and emotional arousal levels. Using advanced sensor data from both the vehicle and the driver, the DrDT simulates, analyzes, and predicts driving behavior to dynamically assess driving fitness. The system architecture uses predefined classes and modules to represent user characteristics and contextual conditions, ensuring personalized and adaptive interactions with applications and devices. The proposed model was validated using the CARLA simulator in conjunction with actual driver data, demonstrating that the personalized FtD system significantly outperforms a generic FtD baseline model in prediction accuracy and consistency. Despite promising results in promoting safer driving practices and reducing accident risks, the research highlights important considerations regarding privacy, security, and ethical implications associated with the deployment of digital twins.