Efficient and validated simulation of crowds for an evacuation assistant

Abstract

To improve safety at mass events, an evacuation assistant that supports security services in case of emergencies is developed. One central aspect is forecasting the emergency egress of large crowds in complex buildings. This requires realistic models of pedestrian dynamics that can be simulated faster than real-time by using methods applied in high performance computing. We give an overview of the project and present the actual results. We also describe the modeling approaches used thereby focusing on the runtime optimization and parallelization concepts.

Problem Decomposition for Evacuation Simulation Using Network Flow

“…In the case that a large group like this forms, the likelihood is that the individuals involved will move slowly, and this may be a fruitful area for exploiting spatial-locality domain subdivision methods such as those suggested in [2,6].…”

“…As a consequence, the complexity of a naive simulation is O(n²) as each occupant checks each other occupant for collisions. EvacSim makes use of a Cell List mechanism [2,21] which limits these collision and interaction checks to locally surrounding occupant agents. Using this mechanism, the computational complexity relates to the density of occupants (i.e.…”

“…Wagoum et. al present a force-based distributed evacuation simulator [2]. This simulator makes use of a 2D vector-force based simulation model similar to the EvacSim simulator used in our work and parallelizes the simulation task via spatial decomposition of the building based on exit placements.…”

“…Simulating the movement of evacuating pedestrians can provide useful information to building designers and to fire safety officials, including evacuation time estimates and locations at risk of traffic congestion or evacuee injury [1][2][3][4][5][6][7]. We make use of a 2D vector-space pedestrian simulator to simulate the individual movements of pedestrians in the building.…”

“…We make use of a 2D vector-space pedestrian simulator to simulate the individual movements of pedestrians in the building. This simulator provides a realistic model of pedestrian movement but is computationally intensive compared to simpler models such as Cellular Automata [2,4,10] or Graph-based simulations [5]. The usefulness of any prediction information is contingent on both the timeliness and the accuracy of the simulation that generated it.…”