Project overview
Our recent VR dyad study decomposed statistically which kinematic channels of an avatar let an observer decode where the avatar’s operator is attending: gross body orientation (chest, shoulders, hips) dominates redundantly, the rotation of the head relative to the torso is the one unique cue, and the tracked eye direction adds nothing. The decomposition is correlational: all channels were always visible. This project delivers the causal test by selectively hiding avatar parts and measuring what decoding accuracy survives.
Project motivation
Statistical variance decomposition cannot rule out that observers would strategically switch to other cues when their preferred cue disappears. Occlusion also matches practice: in real applications avatars are frequently rendered as upper bodies, floating heads, or head-and-hands rigs, and furniture or other users constantly occlude parts of a body. Knowing which renderable subset of the skeleton preserves shared attention tells avatar designers what they can safely omit [1, 2].
Project goals
- Occlusion experiment. Implement avatar conditions in the existing Unity environment, for example: full body, head plus eyes only, torso only (head rigidly coupled), body without head, and full body with a blurred or static face. Dyads perform the established target-decoding task under each condition, within participants.
- Causal cue mapping. Compare decoding error across conditions against the full-body baseline, and test the two central predictions of the parent study: removing the torso should hurt substantially, and removing or freezing the rendered eyes should not hurt at all at shared-space distances.
You will
- Perform a literature review on avatar fidelity and social cue perception
- Implement occlusion/masking conditions in the existing Unity VR environment
- Run a dyad study reusing the existing OptiTrack + eye-tracking pipeline
- Analyze decoding accuracy with mixed-effects models (existing tooling available)
- Summarize your findings in a thesis and present them to an audience
- (Optional) co-write a research paper
You need
- Strong communication skills in English
- Good knowledge of Unity (avatar rendering, shaders a plus)
- Interest in experimental design and statistics
References
- Piumsomboon, T., Lee, G. A., Hart, J. D., Ens, B., Lindeman, R. W., Thomas, B. H., & Billinghurst, M. (2018). Mini-Me: An adaptive avatar for mixed reality remote collaboration. CHI 2018.
- Schwartz, G., et al. (2020). The Eyes Have It: An Integrated Eye and Face Model for Photorealistic Facial Animation. ACM TOG. https://doi.org/10.1145/3386569.3392493
- Loomis, J. M., Kelly, J. W., Pusch, M., Bailenson, J. N., & Beall, A. C. (2008). Psychophysics of perceiving eye-gaze and head direction with peripheral vision. Perception, 37(9). https://doi.org/10.1068/p5896
- The parent study on avatar-mediated shared attention in virtual reality; details are available from the advisor on request.
Keywords: VR, Avatars, Occlusion, Social Cues, Ablation Study