Dynamic Evacuation Route Planning in ComplexBuildings

Evacuating large buildings safely during emergencies requires timely and informed decision-making.In complex environments such as universities or office buildings, individuals often lack access to real-time hazard information, which limits their ability to choose the optimal evacuation route. Thisthesis investigates how a graph-based evacuation system can dynamically determine and presentthe safest path to an exit for each user, based on current hazards in the building.We developed a simulation tool using a graph model of the Niagara building at Malm¨o Uni-versity, where nodes represent rooms and corridors and edges represent walkable paths. Hazardsdynamically modify the graph by disabling affected edges and nodes, and with the help of a Dijk-stra algorithm, recalculate the shortest safe path in real time. An interactive web app allows usersto manually set their position and receive personalized guidance. In addition to simulations, weconducted real-world walkthroughs with evacuation leaders who were unaware of all hazards, tocompare human decision-making under uncertainty with the system’s path recommendations.The results demonstrate significant improvements in evacuation efficiency, with the system re-ducing evacuation times by up to 115% in complex scenarios with multiple hazards. The real-timehazard notifications proved crucial in preventing unnecessary detours, reducing evacuation timesby up to 88.5 seconds in complex scenarios. The implementation shows promising potential forenhancing emergency response capabilities in various building types, with particular benefits forcomplex environments where occupants may be unfamiliar with all possible evacuation routes.