Agent-based Simulation for IoT Facilitated Building Evacuation

The Internet of Things (IoT) has changed our approach to safety systems by connecting sensors and providing real-time data to endangered people in emergency fire situations. In an emergency, evacuating a building in the shortest time possible is essential. Although optimisation methods give a best case scenario, they do not reflect the realism of actual evacuations. We combine a network flow optimisation algorithm with agent based social simulation to provide more realistic evacuation scenarios. Individual occupants are modelled as computational agents in the simulator. Agents are heterogenous and exhibit social attachment to others and vary their speed of movement. A Belief-Desire -Intention (BDI) agent architecture is used to model the cognitive reasoning of individual agents. The work is applied to a real building, equipped with IoT sensors whose data feed into the network flow model. The results show that: the network flow algorithm performs better than the shortest path algorithm and avoids congestion in building bottlenecks; social attachment slows down evacuation; internal adjacently placed doubled doors are better than having a single double width door for increasing evacuation.