Modeling Movement Direction Choice and Collision Avoidance in Agent-Based Model for Pedestrian Flow

Abstract

Agent-based microscopic pedestrian-flow simulation models are promising tools for designers or engineers to evaluate the level of safety or comfort of crowded pedestrian traffic facilities. Existing models tend to simulate movement direction choice behaviors of a virtual agent based on a joint effect of several physical, psychological, and sociological factors dominating the real-world pedestrian walking behaviors. Challenging questions remain for this type of model, including how to control and balance the influences among these behavioral factors and how to naturally avoid collisions without losing the effect of the behavior factors considered. This article presents an improved utility-maximization approach to determine the movement direction of individuals in an agent-based pedestrian-flow simulation model. A new utility function is proposed. An explicit collision detection and avoidance technique is used as a supplementary rule together with the utility maximization method to improve the collision avoidance behaviors in the model. Simulation experiments are carried out for detailed analyses of agent-movement direction-choice behaviors under the influence of utility values and behavioral factors. It is shown that the new utility function can control and balance the influences among the behavioral factors better and avoid unrealistic direction choices. In addition, simulations of intersecting pedestrian flow based a real pedestrian flow experiment are designed, and simulation results are compared with the experiment results. The comparison demonstrates the improvements of using the collision detection and avoidance technique, and shows that well-configured simulations could be close to the experiment both qualitatively and quantitatively.