AI 페르소나 기반 가상 수요와 HILS 실시간 통합을 고려한 수요응답형교통(DRT) 운영 시뮬레이션 프레임워크

초록

In Demand Responsive Transport (DRT) simulations, vehicle dispatching and routing decisions depend strongly on spatiotemporal demand patterns and passenger behavior, making realistic virtual demand generation essential. However, most existing studies rely on simple statistical assumptions and fail to capture individual heterogeneity such as age and personality, while also limiting real-time validation in Human-in-the-Loop Simulation (HILS) environments. This study proposes a two-stage probabilistic mapping model that generates leisure activities and origin–destination pairs from AI personas defined by demographic attributes and five personality traits, and integrates it into a DEVS-based HILS DRT service framework in which virtual and real user requests share the same event structure. Validation using 600 virtual passengers shows strong agreement with the 2024 National Leisure Activity Survey, with Pearson correlation coefficients exceeding 0.80 across age groups. Furthermore, what-if simulations indicate that the proposed DRT service reduces average total travel time

키워드