Agent-based recreation experiences simulation in a 3D virtual environment

© 2013 Dr. Yiqun ChenSimulation of human behaviour involves many subjective often subtle factors such as perceptions and preferences. This is particularly true when trying to replicate visitors’ behaviour in outdoor recreation environments like urban parks and gardens. In order to manage outdoor recreation resources, an effective visitor simulation system can assist managers and planners in their decision-making processes by performing quick tests on various management alternatives. This thesis describes the development of a visitor simulation system for the management of outdoor recreation environments. The system was designed to be capable of reflecting people’s outdoor environment perceptions and landscape preferences. To achieve this goal, four sequential research components were involved: (1) on-site visitor survey and behaviour pattern analysis; (2) simulation platform implementation; (3) visitor agent model design and (4) calibration and evaluation of the developed system. A sound understanding of the interactions between visitors and their outdoor recreation environments is critical to the design of a simulation system. To collect a variety of visitor data on site, an iPhone based mobile survey (called iSurv) was developed and operated in the Australian Garden for 12 months. Analysis of the collected data demonstrated clear visitor behaviour patterns and provided direct guidance to the visitor agent model design. Choosing the simulation platform is the next crucial step. In this research, the simulation system was derived from a commercial 3D game engine. A 3D simulation environment provides agents with virtual vision to support decision-making. By porting real word GIS data into different scenario models and assembling them together, the virtual infrastructure created in a 3D environment formed the base environment for the agent-based simulation system. Visitor agent model design was the next key step. By analysing the general agent design paradigm, pertinent parameters (e.g., demographics, expectations and feelings) and capabilities (e.g., perception, reasoning and action) of agents were identified. The agent behaviour was modelled through a combination of its parameters, capabilities and the environmental variables. Finally, the simulation system was evaluated and high correlations were reported between observed data and simulated outputs. The effectiveness of the developed system was further illustrated by the cogent simulation results for two future scenarios. The research demonstrated the feasibility of creating a visitor simulation system, to a sufficient level of realism, predominantly based on environmental perceptions and landscape preferences factors. The potential of such a tool in outdoor recreation resources management is clear