Experimenting situated cellular agents in indoor scenario: Pedestrian dynamics during lecture hall evacuation

Agent based computational approach can be fruitfully applied to study the behavior and dynamics of crowds as resulting from the interactive behavior of huge numbers of autonomous entities behaving according to an individual or shared goal. This paper focus on the presentation of a model for the simulation of relevant pedestrian dynamics in indoor scenarios with a modeling approach based on a system of situated reactive agents, i.e. Situated Cellular Agents (SCA). The latter are autonomous entities that behave according to a simple model but are able to adapt to different environmental conditions (spatial situatedness) and to share simple social and behavioral rules (obstacle avoidance, flocking). In this paper we present one of the most recent applications of SCA approach to pedestrian dynamics context. The main innovative contribution of this recent work is given by the representation of agent spatial environment as multiple graphs (i.e. multilayered spatial abstractions). This modeling solution has been studied for sake of realism in modeling scenarios where a system of visitors freely move within a highly structured environment (e.g. representing a working space building). Interesting studied behaviors in this type of scenario concern, for instance, the evacuation of a lecture hall in normal, not emergency, conditions (collaboration is assumed within the crowd in this case). The main contribution of this paper is to provide a use case of application of SCA approach that can contribute to the employment of agent-based technology within pedestrian dynamics according to an approach that takes into account the multidisciplinary requirement of this research context and relevance of compliance between modeling and computational approaches. This work is part of an ongoing research project whose goal is the development of decision support systems to design and manage public spaces and events. © 2009 IEEE.