Integrating sensing, routing and timing for indoor evacuation

•We deploy time expanded versions of sensor graph and route graph.•The emergency risk prediction is linked to the time expanded sensor graph.•The risks of sensors at particular times are propagated to vertices and edges in time expanded route graph.•Foresight evacuation route are generated for the real-time minimal risk along the route.

This paper presents an integration of building sensor information with indoor route graphs for the purpose of real-time evacuation planning. This integration allows dynamic risk assessment by prediction, for both the sensor graph and the route graph, and thus enables evacuation route planning aware of predicted risks. The model assumes a dynamic sensor graph according to building monitoring systems and varying sensor states over time. Sensor states are then propagated to a dynamic route graph through the building with regard to passability of edges at time t, using a risk model that quantifies risks of an impact on evacuation. The paper proves the concept in an agent-based simulation, and tests this way whether foresight computed from the risk-aware model improves the evacuation.