Multi-level simulation of the physical, cognitive and social. (April 2022)

• Record Type: Journal Article
• Authors: Bulumulla, Chaminda; Singh, Dhirendra; Padgham, Lin; Chan, Jeffrey
• Is Part Of: Computers, environment and urban systems. Volume 93(2022)
• Publication Date: 2022-04
• Subjects: Multi-level modelling -- Social network diffusion -- Belief-desire-intention agents -- Agent based simulation -- Large-scale evacuations; City planning -- Data processing -- Periodicals; Regional planning -- Data processing -- Periodicals; 303.4834
• DOI: 10.1016/j.compenvurbsys.2021.101756
• Languages: English
• ISSNs: 0198-9715

Abstract

Abstract: Combining multiple models together is increasingly desirable when developing large and complex agent-based simulations. For many social scenarios involving human behaviour, modelling of cognitive agents are required in simulations, with a representation of diffusion processes that capture the spread of various influences (e.g., information, opinions, rumours) in social networks. A diffusion process affect cognitive processes of agents, in turn impacting agent behaviour in the physical simulation as well as in the diffusion process. This paper presents a generic approach of integrating diffusion modelling into an agent based simulation framework that incorporates a cognitive and a physical model. The models individually encapsulate social, cognitive and physical levels of concern, but collectively form a consistent view of the reasoning agent in the integrated simulation. We apply this framework in a large-scale evacuation case study involving more than 35, 000 agents. Results reveal that the addition of the social level can significantly affect evacuation outcomes. The framework is generic, flexible and supports integration of a variety of social, cognitive and physical models. Highlights: We present a multi-level framework that splits agent reasoning into social, cognitive and physical levels. We provide a generic approach that allows cognitive reasoning to impact diffusion processes in a social network. We provide a generic approach that allows diffusion processes to impact agent behaviour in a physical simulation. Inclusion of information diffusion can substantially affect outcomes of a large-scale evacuation simulation. The framework is generic supporting integration of a variety of social, cognitive and physical models.