Wayfinding and Perception Abilities for Pedestrian Simulations. Andresen, E. Ph.D. Thesis, Forschungszentrum Jülich GmbH Zentralbibliothek, Verlag, Jülich, 2018.
Wayfinding and Perception Abilities for Pedestrian Simulations [pdf]Paper  abstract   bibtex   
Computer simulations of pedestrian dynamics are common and reliable tools in order to evaluate safety risks of facilities. However, still many software frameworks for evacuation simulations imply the assumption that all simulated pedestrians are familiar with their environment and therefore take the shortest path to the outside. In fact, the spatial knowledge of people generally varies. Thus, the assumption that all persons of a building possess comprehensive spatial knowledge is a rough approximation of the reality. Especially for simulations in complex buildings the reliability of this approximation is questionable. In order to make simulations of pedestrian dynamics more reliable in this regard, this thesis introduces a new software framework. This framework provides the possibility to predict route choices of a group of people with varying spatial knowledge degrees. Therefore, the framework also considers selected wayfinding strategies that are applied beside the use of spatial memories. These are using signage, using generalized knowledge about the structure of buildings, and search strategies. In addition, three studies have been conducted in order to investigate wayfinding abilities and strategies of people in office buildings and subway stations. The results of the studies are discussed and are used to calibrate and test the models of the new software framework. Finally, the framework is applied to conduct a case study of an evacuation scenario in a subway station. The case study turns out that the egress time in the station is strongly dependent on the wayfinding strategies and abilities of the occupants. This outcome suggests that the proper consideration and prediction of route choices is relevant and necessary for reliable evacuation simulations

Downloads: 0