Please use this identifier to cite or link to this item:
|Title:||Behavioural modelling of crowds using DI-guy||Authors:||Tan, Daniel Yi Wen.||Keywords:||DRNTU::Engineering||Issue Date:||2013||Abstract:||Crowd is a highly sophisticated social phenomenon and simulating crowd has always been a challenging task in terms of achieving efficiency and realistic crowd behaviour. In order to address this challenging task, an agent-based modelling methodology to populate crowd in order for each individual agent to have their own decision-making ability to simulate crowds in the real world was adopted. The purpose of this project is to investigate on existing frameworks used in agent-based modelling and to create an architecture that simulate closely to human behaviour of decision-making, as well as the capability to perform efficiently in the virtual environment. Through the investigation on two existing frameworks – the Finite State Machines and Belief-Desire-Intention frameworks, a better understanding of the application of the frameworks for different scenarios as well as their advantages and disadvantages were obtained from this study. Finally, a new integrated architecture that uses both of the frameworks was proposed in this paper. It was developed for the agent-based modelling with the use of commercial human simulation software – the DI-Guy. The integrated agent was used to interact with the agent designed using Hierarchical Finite State Machine in DI-Guy in a scenario created to observe the difference in behaviour so as to determine which type of agents will simulate more closely to how a human behaves and react to different situations. In conclusion, the integrated architecture of the agent developed in this paper reacted well in the scenario created, however more research and further improvement are needed in this project to better simulate human behaviour closely. The main recipe of how this integrated agent performs depends on the plan library and a more precise plan has to be designed to meet the dynamic behaviour of human reaction. Crowd simulation with the introduced framework in this study will still need to be tested in order to testify realistic crowd behaviour for future study of crowd behaviour.||URI:||http://hdl.handle.net/10356/52805||Rights:||Nanyang Technological University||Fulltext Permission:||restricted||Fulltext Availability:||With Fulltext|
|Appears in Collections:||SCSE Student Reports (FYP/IA/PA/PI)|
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.