Embodiment in Virtual Environments - Analysing the Influence of Avatar Representation and Latency

This thesis deals with the influence of latency and avatar representation on the ability to accept a virtual avatar as your own body.
A study was designed and conducted in which subjects performed several tasks while being exposed to scenarios of different latencies and avatar representations.
Our results show that the quality of the avatar has a higher impact on perceived embodiment of a user than previously expected.

Description

The user's immersion in virtual environments depends heavily on whether they feel represented by the avatar in the virtual reality and whether they have the feeling that they are the avatar themselves.
This feeling is called embodiment in research. Embodiment is composed of three elements: Body Ownership, Agency and Self-Location. Two important factors that influence embodiment are latency and avatar representation.

The goal of the thesis was to analyse the influence of latency and avatar representation on the ability to accept a virtual avatar as your own body. In order to analyze these two factors, a study was designed and conducted. The Unreal Engine project implemented for the study allowed subjects to choose between two avatars that varied in terms of their creation and quality. The mesh avatar was not customized to match the user's appearance and was controlled by five position sensors. In contrast, the point cloud avatar is created through a capture of an RGB depth camera and adapts its position, movement and appearance based on the user's captured image. Compared to the mesh avatar, the point cloud avatar is of higher quality. Three delays between the movement of the subject and the movement of the avatar could be set in the software. The delay between the movement of the subject and the movement of the avatar was 150 ms, 225 ms or 300 ms. The subjects completed three different tasks and answered a 7-point Likert scale questionnaire.

Results

The results of the study suggest that the avatar representation has a greater impact on the subjects' embodiment than the level of latency. The embodiment was worse when using a lower-quality avatar than when the latency was increased by 75 ms. Even at latencies of 300 ms, a point-cloud avatar evoked stronger embodiment than a mesh avatar at a latency of 150 ms.

In contrast to the mesh avatar, the user's embodiment does not collapse when he is utilizing a point-cloud avatar with a perceived latency of 300 ms. The selection of a point cloud avatar results in much higher ratings for body ownership, agency, and self-location on the questionnaire. The rise in latency only leads to significantly worse body ownership and agency when a subject is using a mesh avatar.

Furthermore, for the point-cloud avatar, the threshold at which the user perceived a decreasing synchronization between his own movements and the avatar's movements was 75 ms higher compared to the threshold for the mesh avatar.

Based on the evaluation of the results, it is recommended to use higher quality avatars, even if the user is exposed to increased latency as a result.

Files

Full version of the bachelor thesis (German only)

This video shows a summary of the tasks completed by the subjects:

These videos show a whole run of all tasks completed by the subjects. The subject used a mesh avatar in the left video and a point cloud avatar in the right one:

License

This original work is copyright by University of Bremen.
Any software of this work is covered by the European Union Public Licence v1.2. To view a copy of this license, visit eur-lex.europa.eu.
The Thesis provided above (as PDF file) is licensed under Attribution-NonCommercial-NoDerivatives 4.0 International.
Any other assets (3D models, movies, documents, etc.) are covered by the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. To view a copy of this license, visit creativecommons.org.
If you use any of the assets or software to produce a publication, then you must give credit and put a reference in your publication.
If you would like to use our software in proprietary software, you can obtain an exception from the above license (aka. dual licensing). Please contact zach at cs.uni-bremen dot de.