Vision-based crowd dynamic model for multi-directional pedestrian flow at public space: a hybrid approach

Abstract

Visual information perceived by pedestrians plays a vital role in tactical level decision-making and changes continuously during the movement. Changes in the pedestrians' position also synchronously update the content of alternative paths from their location to the destination in a public space with internal obstacles and dynamic pedestrian flow. In this paper, we propose a microscopic pedestrian simulation model with continuous-space representation for multi-directional flow analysis. Assuming that a pedestrian can only perceive the congestion level and the choices of other pedestrians within its visual field, the route choice decision of the pedestrian around obstacles is formulated. In addition, integrated network representation enables the modeling of a pedestrian's route-planning behavior with an optimal reciprocal collision avoidance (ORCA) system. Further, numerical simulations verified the ability of the model to reproduce the behavior observed in the intersection zone. This study can benefit the pedestrian flow arrangement and structural design in large public spaces.