Supporting the Design, Selection, and Evaluation of Accessible Interaction Techniques for Virtual Reality

Abstract

Virtual Reality (VR) is rapidly gaining popularity as a consumer technology, yet accessibility has not been prioritized. VR enables users to interact with immersive virtual worlds, but effective interaction requires the ability to perceive and navigate the environment. Fortunately, over the past thirty years, hundreds of scene-viewing and locomotion techniques that facilitate viewing and navigating in VR environments have been developed that could potentially be accessible to individuals with various physical impairments. However, it remains unclear which techniques are optimal for specific impairments, and the process of designing accessible techniques is largely unexplored. The goal of my dissertation is to provide researchers and designers with guidance on selecting or designing accessible scene-viewing and locomotion techniques to enhance VR accessibility for individuals with upper-body impairments.To achieve this goal, I developed a taxonomy and design framework to help designers select existing scene-viewing techniques that require minimal head movement as well as create new scene-viewing techniques based on user and virtual environment (VE) characteristics. Additionally, I created an extensible testbed as a step towards standardizing the evaluation of VR interaction techniques. Using this testbed, I evaluated six locomotion techniques with participants both with and without upper-body impairments and found that Teleport was the most accessible technique. I also found that people with and without impairments performed similarly with Sliding Looking. By analyzing low-level headset and controller data collected during the study, I examined differences in how individuals with and without impairments moved and interacted with the controllers to gain insight into how locomotion technique design impacted performance for both groups. I found that headset metrics differentiated groups across all six techniques and that movement-, button-, and target-related metrics could explain performance differences between groups. Finally, I developed a user model capable of predicting the most suitable locomotion techniques for individual users based on their abilities. The model could predict a user’s fastest locomotion technique with 85% and accuracy and predicted times were within 16% of actual times it took participants to use the fastest technique. The thesis this work will demonstrate is: The accessibility of virtual reality (VR) can be improved by the selection, design, and evaluation of scene-viewing and locomotion techniques using design frameworks, testbed evaluations, and movement analysis and modeling.