TY - JOUR AU - Xia, Long AU - Song, Weiguo PY - 2022/01/19 Y2 - 2024/03/29 TI - A Memory-Based Evacuation Navigation Model in Complex High-Rise Buildings JF - Collective Dynamics JA - Coll Dyn VL - 6 IS - 0 SE - Pedestrian and Evacuation Dynamics 2021 DO - 10.17815/CD.2021.123 UR - https://collective-dynamics.eu/index.php/cod/article/view/A123 SP - 1-16 AB - In contemporary society, safety issues are the main focus in the field of pedestrian and evacuation dynamics. As for complex high-rise buildings, the navigation strategies of evacuees still need to be further studied. Previous types of research has contributed to the construction of evacuation navigation model in complex high-rise buildings, where pedestrians are regarded as having an omniscient view in most of these models. In reality, evacuees’ perception is always limited, especially when the scenario is complex. In this contribution, pedestrians’ perception procedure is considered by computing the visible space so that the occlusion of the visual field can be estimated. In addition, human memory progress is modeled. Not all parts of environmental information would be remembered. Driven by evacuees’ memory data, a proposed dynamical shortest path algorithm will be periodically implemented or suddenly triggered by incidents during the simulation. For the pedestrians who have no enough knowledge about the evacuation scenario, a communication system is utilized so that information can be obtained from well-knowledged pedestrian, and an autonomous way-finding system will be executed when useful information cannot be acquired through near evacuees. For the microscopic perspective, human following and avoidance behavior is modeled. Simulations in two different types of scenarios are conducted. The knowledge level of simulated agents is gradually evolved by in-room free exploration. Results in different conditions show that the proposed memory-based model can reproduce pedestrians’ observation, turn-back, communication, and searching behavior. The intense conflict caused by the bidirectional crowd is observed and analyzed. In addition, the effects of knowledge level are investigated. The presented model in this contribution can be promising and useful in safety engineering. ER -