Screening Urban Structural Vulnerability for Future Robot–Pedestrian Coexistence : An Exploratory Pandemic-Era Delivery-Pressure Scenario

다국어 초록 (Multilingual Abstract)

This study proposes an exploratory stress-testing framework for screening urban structural vulnerability prior to the deployment of sidewalk autonomous delivery robots. Because empirical robot–pedestrian interaction data remain limited, the COVID-19 pandemic is treated as a historical urban scenario characterized by concentrated delivery demand. The intensity of food-and-beverage delivery demand represents delivery pressure on urban space, while Emergency Medical Services (EMS) incident records provide observed incident outcomes involving vulnerable pedestrians during the same period. The comparison between pandemic-era two-wheeled delivery activity and future autonomous delivery robot operations is restricted to the spatiotemporal concentration of demand and persistent built-environment constraints, rather than vehicle behavior or accident outcomes. XGBoost and SHAP analyses identified nonlinear predictive associations among delivery pressure, vulnerable -pedestrian exposure, slope, road width, and related interaction features. K-means clustering and micro-level Street View audits were used to examine the spatial typologies and persistent physical constraints of high-vulnerability areas. Based on these findings, the study proposes time-sensitive operational management and adaptive street-design strategies. The resulting index is intended to prioritize locations for future robot field trials and urban-design audits, providing a practical framework for data-driven and inclusive smart-city planning.