Fuzzy Cellular Automata Models for Crowd Movement Dynamics at Signalized Pedestrian Crossings

Abstract

Crowd movement dynamics at a signalized pedestrian crossing constitutes a complex system affected by many factors. Existing crowd simulation models seldom consider cognition and decision making of individual pedestrians. In this study, a fuzzy cellular automata (FCA) model was developed to simulate pedestrian movements at crowded signalized pedestrian crossings that incorporated pedestrian decision-making processes. The proposed FCA model incorporated fuzzy logic into a conventional cellular automata (CA) model. In contrast to existing models and applications, the proposed FCA model used fuzzy sets and membership functions to simulate individuals’ decision making process. Four fuzzy sets were applied for each decision: stop or go (stop–go) decision, moving direction, velocity change, and congregation. Membership functions of each input factor as well as weight factors of each fuzzy set at different movement zones were calibrated on the basis of field observations. Model performance was assessed by comparisons of trajectories between estimation and observation, interactions with conflicting vehicles and pedestrians, and congregation phenomena. Through a simulation experiment and comparison with existing approaches, simulation results show that the FCA model can well replicate crowd movement dynamics in real-world pedestrian crossings.