Room evacuation in the presence of obstacles using an agent-based model with turning behavior

Pedestrian evacuation in a room environment is critical to architectural design, safety management and code development. The complexity of this issue is further increased in the presence of multiple obstacles, which may affect pedestrian dynamics and evacuation pattern. Most of previous studies on the room evacuation in the presence of multiple obstacles do not account for pedestrian's turning or rotational behavior, this paper aims to address this limitation. For this purpose, a previously developed agent-based model considering agent's turning behavior is extended in the context of room evacuation. Several improvements are made to this model, including: a procedure for generating feasible evacuation routes based on concepts of visible regions and accessible nodes; a deterministic approach based on the estimated evacuation time to select final evacuation route from feasible ones; and reformulation of agent's attractive potential to consider its movement along evacuation route. Simulation results with the proposed agentbased model are compared with published experimental data, and reasonable agreement can be observed. It is concluded that the proposed agent-based model is suitable for studying room evacuation problem in the presence of multiple obstacles under relatively low density scenario.