A Configurable Agent-Based Crowd Model with Generic Behavior Effect Representation Mechanism

Abstract

Most of the existing crowd models were designed for specific behaviors or scenarios (e.g., emergency evacuations and bi-directional crowd flows). Significant modifications were often required to accommodate new behaviors or new scenarios. This article proposed a generic crowd model with the flexibility to incorporate different behaviors under different scenarios. At the higher level of the proposed crowd model, the agent-based modeling method was used to enable the individual heterogeneity and decision making. At the lower level, a unified mechanism to represent the effects of different individual behaviors was introduced. A core formula with seven generic parameters (i.e., agent's position, target's position, behavior angle, effect of base speed, agent factor, target factor, and distance factor) has been developed to form the basis of the unified mechanism. This article also presented a Behavior Library that consisted of a set of basic behaviors that were able to construct complex behaviors through their combinations. To demonstrate the capability of the model in various scenarios, the following simulations have been implemented and discussed: queuing at an exit, bi-directional pedestrian walk flow, evacuation in a building, and consensus decision making in a large group. © 2014 Computer-Aided Civil and Infrastructure Engineering.