Integrating Machine Learning and Smart Wayfinding Solutions for Riverine Flood Emergency and Evacuation Management in Urban Built Environments
Year: 2026
Type: conference-paper
Abstract: Floods are the most destructive sudden-onset disasters worldwide, affecting more people annually than any other natural hazard, especially in urban areas. Riverine floods surely represent critical phenomena in this context, also due to the interactions between the hazard source and additional boundary conditions (especially for coastal cities), the territorial and urban physical vulnerability, and the “user factors”. These components shape how flood events impact the urban built environment and its population. In some cases, emergency and evacuation strategies could also be activated to mitigate effects on exposed users. Ensuring timely and reliable management solutions and strategies is therefore essential, especially given the increasing complexity of the dynamics of climate-driven floods and population/visitors' flows. To support these goals, this work provides an integrated, data-driven system to support alert and outdoor evacuation during riverine flood emergencies in urban scenarios. The system first innovatively incorporates a machine-learning alert module, which relies on historical data of previous flood events and continuously analyses sensor data to predict hazardous outdoor conditions, enabling early and targeted warnings for the population. When a possible flood emergency scenario is detected, the system defines danger zones, a dynamic routing module identifies proper evacuation paths, and addresses direct wayfinding towards citizens, using communication via mobile devices. Emergency operators access a dedicated dashboard for real-time monitoring, route management and validation. Developed within the CLIMRES project, the system is showcased considering the pilot scenario of the city centre of Senigallia, Italy, a coastal city affected by riverine flood events and a relevant tourist destination along the Italian Adriatic Coast. The combination of predictive analytics with adaptive wayfinding enhances situational awareness and supports fast, coordinated and context-sensitive emergency and evacuation management.
Institution: Marche Polytechnic University
Language: English
Subfield: Global and Planetary Change
Field: Environmental Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7204185203