M.S. (Master of Science)
Department of Mechanical Engineering
Mechanical engineering; Behaviorism (Psychology)
Humans are social by nature, and therefore when they come together in a large group, the resulting crowd behavior is difficult to explain in terms of individual motion dynamics. In pedestrian crowds, social influence occurs in the form of collision avoidance and in following neighbors' actions. Understanding the dynamics of social influence can help improve crowd management strategies. Obtaining experimental data with crowds, however, is impractical due to the number of subjects needed and unsafe due to possible hazards of having many people assemble in a single place. The goal of this research is to develop a realistic virtual reality platform for crowd behavior studies and use it to validate the effect of social influence in human crowds. The specific objective of this research is to use methods from dynamical systems modeling to enable a custom virtual environment of an interactive humanoid crowd. The virtual environment is then evaluated through a behavioral experiment. Specifically, in this thesis we first adapt pedestrian motion models from the literature to produce dense pedestrian crowds where individuals can maintain close proximity and avoid unrealistic turning motions. We modify the collision avoidance and interaction functions so that the computational burden is reduced to realistically simulate human crowds of up to sixty individuals at real-time speeds on a 3D graphics environment. The crowd simulation is then utilized to test the hypothesis that behavioral contagion triggered in the form of a subset of virtual characters shifting their gaze upwards can induce a human participant to follow that action. A human-subjects study with seventy participants is conducted to assess the performance of the virtual environment as well as test the hypothesis. Participants are surveyed on the believability of the virtual environment and their postural responses measured to record the effect of behavioral contagion. Our results show that the virtual environment is effective in being perceived as natural and realistic and that participants are able to interact with the virtual crowd without experiencing any visual delays. We find that participants follow the visual gaze of virtual characters and that this action is proportional to the number of characters that trigger such a gaze. Contagion effect is further established in terms of the time spent looking up by the participants which is also found to be dependent on the size of the stimulus group.
Mohammadi Jorjafki, Elham, "Modeling and validation of dynamics of social influence in virtual human crowds" (2018). Graduate Research Theses & Dissertations. 3992.
Northern Illinois University
Rights Statement 2
NIU theses are protected by copyright. They may be viewed from Huskie Commons for any purpose, but reproduction or distribution in any format is prohibited without the written permission of the authors.
Advisors: Sachit Butail.||Committee members: Sachit Butail; Ji-Chul Ryu; Brad Sagarin.||Includes illustrations.||Includes bibliographical references.