Knowledge-Based Crowd Motion for the Unfamiliar Environment

Abstract

This paper presents a knowledge-based framework for modeling crowd motion in an unfamiliar environment. Three predominant procedures were defined in the framework: knowledge representation, dynamic knowledge transmission, and knowledge-guided wayfinding. First, a semantic-based layered structure is designed to represent knowledge about the unfamiliar environment. Second, to capture important influential factors for knowledge transmission, such as personal influential radius and personal abilities of expressing and assimilating knowledge, we construct a model of personalized knowledge transmission. Third, a probability-based knowledge-guided wayfinding model is presented to produce diverse actions and allow the individuals to adapt to knowledge changes in an unfamiliar environment. Finally, a crowd simulation system is implemented to visualize the analysis in a graphical manner. The proposed method is expected to provide guidance for emergency management especially when the crowd has incomplete knowledge of the environment.