Presenter/Author Information

Stefania Bandini
Sara Manzoni
G. Vizzari

Keywords

multi-agent system, simulation, crowd behaviour

Start Date

1-7-2004 12:00 AM

Abstract

The paper presents a Multi Agent Systems (MAS) approach to crowd modelling, based on the Situated Cellular Agents (SCA) model. This is a special class of Multilayered Multi Agent Situated System (MMASS), rooted on Cellular Automata, providing an explicit spatial representation and the definition of adjacency geometries. The model also defines a concept of autonomous agent, provided with an internal architecture and individual state and behaviour, capable of different means of space–mediated interaction (synchronous, between adjacent agents, and asynchronous among distant entities). Heterogenous entities may be modelled through the specification of different agent types, defining different behaviours and perceptive capabilities. After a brief description of the model, its application to simple crowd behaviours will be given (e.g. lane and group formation), and an application providing the integration of a bidimensional simulator based on this model and a 3D modelling application (3D Studio) will also be described. The adoption of this kind of system allows to specify, simulate and evaluate a design solution, but also to easily produce a realistic visualization of the simulation, in order to facilitate the communication with involved actors. In fact, while expert decision– makers often require only abstract and analytical results deriving from the simulation, other people involved in the decision–making process related to the design may be helped by other forms of graphical representation.

COinS
 
Jul 1st, 12:00 AM

Situated Cellular Agents for Crowd Simulation and Visualization

The paper presents a Multi Agent Systems (MAS) approach to crowd modelling, based on the Situated Cellular Agents (SCA) model. This is a special class of Multilayered Multi Agent Situated System (MMASS), rooted on Cellular Automata, providing an explicit spatial representation and the definition of adjacency geometries. The model also defines a concept of autonomous agent, provided with an internal architecture and individual state and behaviour, capable of different means of space–mediated interaction (synchronous, between adjacent agents, and asynchronous among distant entities). Heterogenous entities may be modelled through the specification of different agent types, defining different behaviours and perceptive capabilities. After a brief description of the model, its application to simple crowd behaviours will be given (e.g. lane and group formation), and an application providing the integration of a bidimensional simulator based on this model and a 3D modelling application (3D Studio) will also be described. The adoption of this kind of system allows to specify, simulate and evaluate a design solution, but also to easily produce a realistic visualization of the simulation, in order to facilitate the communication with involved actors. In fact, while expert decision– makers often require only abstract and analytical results deriving from the simulation, other people involved in the decision–making process related to the design may be helped by other forms of graphical representation.