A Hybrid Crowd Simulation Framework Towards Modeling Behavior of Individual Avoidance of Crowds
Loading...
Date
2015
Journal Title
Journal ISSN
Volume Title
Publisher
ACM Siggraph
Abstract
Crowd movement is a common but complicated phenomenon in our daily lives. The behaviors of crowds can be affected by both individual and crowd. Most previous research could be categorized as either agent-based methods [van den Berg et al. 2009], which have advantage on simulating individual behaviors, or continuous methods [Narain et al. 2009] which are efficient for simulating crowds with large population. To take advantage of both, [Golas et al. 2013] proposed a hybrid solution which combined and blended both methods. [Bruneau et al. 2015] proposed a virtual reality based study to measure the behavior of individual avoidance of crowds. Their study showed that people have different choices of going through or around based on the density, moving direction and type of crowd. During their experiments, they found significant individual difference between subjects, but not studied as a factor. In this poster, we focus on simulating individual differences on choices of going through or avoiding crowds. We introduce an empirical agent model for individual avoidance behaviors. By integrating personality trait into our agent model, we are able to simulate individual difference of avoid or join behavior. We also present our hybrid crowd simulation framework which can automatically identify individuals and crowds, and explicitly trigger individual avoidance of crowds during simulation.
Description
@inproceedings{10.1145:2786784.2795135,
booktitle = {ACM/ Eurographics Symposium on Computer Animation},
editor = {Florence Bertails-Descoubes and Stelian Coros and Shinjiro Sueda},
title = {{A Hybrid Crowd Simulation Framework Towards Modeling Behavior of Individual Avoidance of Crowds}},
author = {Liu, Haiying and Yan, Zhixin and Lindeman, Robert W. and Ding, Gangyi and Huang, Tianyu},
year = {2015},
publisher = {ACM Siggraph},
ISBN = {978-1-4503-3496-9},
DOI = {10.1145/2786784.2795135}
}