Help Me Evacuate! A Smart Adaptable Evacuation System With Congestion Prediction Capabilities

Abstract

Emergency evacuation planning is a vital problem that affects building occupants' safety. The commonly used static evacuation plans rely on static signs disregarding crowd density changes. Such static plans often lead to congestion at emergency exits; since many occupants tend to avoid following exit signs as they feel safer following the crowds exiting the building or following other paths familiar to them. This paper proposes a smart and adaptable evacuation system that predicts congestion and adapts accordingly to minimize evacuation time. We introduce a simulation model that mimics occupants' movement in different building layouts. The proposed system performs Monte Carlo simulations to forecast possible congestion locations and guide occupants away from them. Guiding directions are displayed and updated to consider dynamic environment changes. We evaluated our approach compared to a greedy evacuation method that relies on static exit signs, showing a significant evacuation time improvement of 21% achieved on average by our approach.