Virtual Avatars

• Group Student Project for an Intro to Human-Computer Interaction course at University of Delaware

  • Creating an application in Unity (including a 3D virtual avatar and an interactive 3D virtual environment)

  • Researching, storyboarding, prototyping and presenting deliverables 

There is a growing deficiency in understanding the perspectives of people of different races/ethnicities.

As we re-acclimate to social interactions following the pandemic and other issues of racial injustices, we must be aware of how our actions may affect those around us.

We need practice mitigating implicit racial biases in different social environments.

The initial goal of the virtual avatars project was to create an application that allowed users to interact with a virtual avatar in a virtual environment simulating real-world situations.  A user would be able to practice how to deal with certain situations that may propel or lessen implicit biases depending on the user's actions.

Previous Research

• One experiment demonstrated that embodying light-skinned users in a dark-skinned virtual avatar significantly decreased implicit racial bias against dark-skinned people, when compared to embodying light-skinned, purple-skinned, or no virtual avatars. [1]

• Other research shows that virtual environments can influence users' feelings and how they perceive virtual avatars. [2]

Pre-Survey

This Google Forms survey (around 7 - 10 questions total) was intended to be taken by participants before they used our application.  It asks questions about participants' backgrounds, experiences with implicit biases and gauges their familiarity with AR or VR experiences. Some example questions are shown below:

Storyboarding

We storyboarded our project using Miro. Our prototype mainly focused on showing the user interacting with the virtual avatar in a virtual environment.

Hardware

Our initial goal was to use either Oculus VR or Windows Mixed Reality headsets to allow our users to interact with the virtual environments. Our final product ended up not implementing VR, and was instead presented as a Windows 64-bit system executable application.

Software

We created our final application in Unity, using packages recommended to us by our mentor.  We used Adobe Mixamo to get a base virtual avatar which we then customized in Unity.  We also wrote C# scripts in Visual Studio Code to implement audio capabilities in our virtual avatar.

Setbacks

We planned to develop our application to be implemented through VR, then AR, and finally we settled with developing for a Windows executable due to time constraints, learning curves and limited resources.  Our team's individual schedules made it difficult to collectively meet at the HCI Lab, where we only had one machine available to work on our project & perform testing.  At one point in development, there was also an issue with a Unity update erasing some of our previous work that allowed for virtual avatar facial movements.  To combat these setbacks, we had to schedule additional times to meet in the HCI Lab outside of our usual schedules, which our instructors and mentors supported us in.  Finally, we were able to regain most of our lost progress, but due to time constraints, our final application did end up not having facial movements to match the audio of our virtual avatar's speech.

Conclusions & Future Work

Despite these challenges, we managed to deliver a working product that allows users to interact with a virtual avatar in a virtual environment, which is at the core of what we set out to do.  We also gained a lot of experience working with Unity as complete novices, and by the end of the project, we all learned how to create virtual avatars, create virtual environments, customize virtual avatars and add speech and facial movement capabilities to virtual avatars within Unity.

In the future, we would like our project to further explore mitigating implicit biases by adding in different virtual environments to simulate various social situations.  Another goal would be to implement VR or AR capabilities as we initially planned in this project, although in its current state, VR would be the easiest to implement of the two. 

[1] Tabitha C. Peck, Sofia Seinfeld, Salvatore M. Aglioti, Mel Slater, Putting yourself in the skin of a black avatar reduces implicit racial bias, Consciousness and Cognition, Volume 22, Issue 3, 2013, Pages 779-787, ISSN 1053-8100, https://doi.org/10.1016/j.concog.2013.04.016.

[2] Katharina Legde and Douglas W. Cunningham. 2019. Evaluating the Effect of Clothing and Environment on the perceived Personality of Virtual Avatars. In <i>Proceedings of the 19th ACM International Conference on Intelligent Virtual Agents</i> (<i>IVA '19</i>). Association for Computing Machinery, New York, NY, USA, 206–208. DOI:https://doi.org/10.1145/3308532.3329425