The Effect of Anthropometric Properties of Self-Avatars on Action Capabilities in Virtual Reality
Date of Award
Doctor of Philosophy (PhD)
School of Computing
Sabarish V Babu
Christopher C Pagano
Larry F Hodges
The field of Virtual Reality (VR) has seen a steady exponential uptake in the last decade and is being continuously incorporated into areas of popular interest like healthcare, training, recreation and gaming. This steady upward trend and prolonged popularity has resulted in numerous extravagant virtual environments, some that aim to mimic real-life experiences like combat training, while others intend to provide unique experiences that may otherwise be difficult to recreate like flying over ancient Egypt as a bird. These experiences often showcase highly realistic graphics, intuitive interactions and unique avatar embodiment scenarios with the help of various tracking sensors, high definition graphic displays, sound systems, etc. The literature suggests that estimates and affordance judgments in VR scenarios such as the ones described above are affected by the properties and the nature of the avatar embodied by the user. Therefore, to provide users with the finest experiences it is crucial to understand the interaction between the embodied self and the action capabilities afforded by it in the surrounding virtual environment.
In a series of studies aimed at exploring the effect of gender matched body-scaled self-avatars on the user's perception, we investigate the effect of self-avatars on the perception of size of objects in an immersive virtual environment (IVE) and how this perception affects the actions one can perform as compared to the real world. In the process, we make use of newer tracking technology and graphic displays to investigate the perceived differences between real world environments and their virtual counterparts to understand how the spatial properties of the environment and the embodied self-avatars affect affordances by means of passability judgments. We describe techniques for creation and mapping VR environments onto their real world counterparts and the creation of gender matched body-scaled self-avatars that provides real time full-body tracking.
The first two studies investigate how newer graphical displays and off-the-shelf tracking devices can be utilized to create salient gender matched body-scaled self-avatars and their effect on the judgment of passability as a result of the embodied body schema. The study involves creating complex scripts that automate the process of mapping virtual worlds onto their real world counterparts within a 1cm margin of error and the creation of self-avatars that match height, limb proportions and shoulder width of the participant using tracking sensors. The experiment involves making judgments about the passability of an adjustable doorway in the real world and in a virtual to-scale replica of the real world environment. The results demonstrated that the perception of affordances in IVEs is comparable to the real world but the behavior leading to it differs in VR. Also, the body-scaled self-avatars generated provide salient information yielding performance similar to the real world. Several insights and guidelines related to creating veridical virtual environments and realistic self-avatars were achieved from this effort.
The third study investigates how the presence of body-scaled self-avatars affects the perception of size of virtual handheld objects and the influence of the person-plus-virtual-object system created by lifting the said virtual object on passability. This is crucial to understand as VR simulations now often utilize self-avatars that carry objects while maneuvering through the environment. How they interact with these handheld objects can influence what they do in critical scenarios where split second decisions can change the outcome like combat training, role-playing games, first person shooting, thrilling rides, physiotherapy, etc. It has also been reported that the avatar itself can influence the perception of size of virtual objects, in turn influencing action capabilities. There is ample research on different interaction techniques to manipulate objects in a virtual world but the question about how the objects affect our action capabilities upon interaction remains unanswered, especially when the haptic feedback associated with holding a real object is mismatched or missing. The study investigates this phenomenon by having participants interact with virtual objects of different sizes and making frontal and lateral passability judgments to an adjustable aperture similar to the first experiment. The results suggest that the presence of self-avatars significantly affects affordance judgments. Interestingly, frontal and lateral judgments in IVEs seem to similar unlike the real world.
Investigating the concept of embodied body schema and its influence on action-capabilities further, the fourth study looks at how embodying self-avatars that may vary slightly from your real world body affect performance and behavior in dynamic affordance scenarios. In this particular study, we change the eye height of the participants in the presence or absence of self-avatars that are either bigger, smaller or the same size as the participant. We then investigate how this change in eye height and anthropometric properties of the self-avatar affects their judgments when crossing streets with oncoming traffic in virtual reality. We also evaluate any changes in the perceived walking speed as a result of embodying altered self-avatars. The findings suggest that the presence of self-avatars results in safer crossing behavior, however scaling the eye height or the avatar does not seem to affect the perceived walking speed. A detailed discussion on all the findings can be found in the manuscript.
Bhargava, Ayush, "The Effect of Anthropometric Properties of Self-Avatars on Action Capabilities in Virtual Reality" (2019). All Dissertations. 2502.