Crowd simulation is a crucial tool for architects and planners to effectively manage large gath- erings and prevent congestion. Accurate predictions of crowd behavior, especially in cross-flow scenarios where multiple streams intersect, are essential to avoiding high-density situations that can lead to fatalities. Modeling counter-flowing crowds is a significant challenge due to the com- plex decision-making processes individuals employ when reacting to external stimuli. Traditional models often simplify these behaviors using attractive and repulsive forces, which fail to fully recreate the complexity of real-world scenarios. To address these challenges, more sophisticated modeling approaches have been developed, such as "switching behavior" ("Modeling of Behavioral Changes in Agent-Based Simulations", Dr. Benedikt Kleinmeier, 2021) and "time-to-collision" ("Body and Mind", Iñaki Echeverría- Huarte, Alexandre Nicolas, 2023), which aim to enhance the accuracy of counter-flow modeling. This thesis examines and compares these approaches by implementing their counter-flow solution strategies into crowd:it, an agent-based microscopic crowd simulation software. The approaches are going to be compared and evaluated by quantitative metrics. Quan- titative metrics include time to exit counter flow area, density and velocity development via heatmaps, lane formation ("Simulating pedestrian dynamics", Dr. Michael Seitz, 2016) and tra- jectory comparison with experimental data from Forschungszentrum Jülich. This research seeks to improve the accuracy and enhance applicability to complex scenarios of crowd simulations, thereby enhancing safety and planning at large-scale events.
Aufgabensteller:
Prof. Dr. D. Kranzlmüller
Dauer der Arbeit:
Anzahl Bearbeiter: 1
Betreuer: