A generic approach to
modelling individual behaviours in crowd simulation

Sun, Q

Abstract

Crowd simulation has been widely used to simulate crowd dynamics and their behaviours. However, majority of existing studies can only simulate a specific scenario or behaviour. Although recent developments have attempted to integrate different individual behaviours in order to achieve a more realistic simulation result, it is still very complex and those crowd models often require significant modifications.

This study is therefore aimed to develop a generic crowd model, which provides the flexibility to configure and represent different scenarios, as well as the ability to demonstrate individual differences on crowd behaviours. The theoretical principle of the proposed crowd model is based on the combination of force-based modelling and agent-based modelling. A unified core mathematical formula, which contains seven key parameters, is developed to represent the generic behaviour effects. In addition, a Behaviour Library is developed to present a set of basic behaviours by using the unified formula and subsequently, more complex behaviours could be formed by combining the basic behaviours. The proposed crowd model is implemented in a simulation environment by using Microsoft XNA framework. A number of well-known crowd behaviours are tested with the crowd model for validation. The proposed crowd model is further validated by simulating real life experiments and comparing its results.

This research study presents a novel approach to simulate crowd behaviour at individual level by introducing a generic crowd model that can be configured into specific scenarios. It introduces a theoretical concept, through which different behaviour effects could be quantified by a unified mathematical formula. As a result, crowd modelling and simulation of different scenarios can be significantly simplified. For future work, the proposed crowd model can be tested under complex environment in order to fine-tune its theoretical model and to expand the Behaviour Library.

Citation

modelling individual behaviours in crowd simulation. (Thesis). University of Salford

• Thesis Type: Thesis
• Deposit Date: Mar 7, 2014
• Publicly Available Date: Apr 5, 2016