Supported navigation in immersive virtual environments

• Unterstützte Navigation in Immersiven Virtuellen Umgebungen

Freitag, Sebastian; Kuhlen, Torsten (Thesis advisor); Steinicke, Frank (Thesis advisor)

Aachen (2018)
Dissertation / PhD Thesis

Dissertation, Rheinisch-Westfälische Technische Hochschule Aachen, 2018

Abstract

Navigation - the process of getting from one location in a virtual scene to another - is one of the most important and fundamental types of interaction in Virtual Reality, as it is part of most use cases and applications in some way or another. However, there is no single navigation technique that is suitable and appropriate for all use cases, as there is typically a different set of requirements for each application, e.g., regarding realism, efficiency, cognitive load, cybersickness, etc. Furthermore, all techniques come with their own set of drawbacks. For example, while mapping the user's physical motion directly to their virtual motion ("real walking") is the most intuitive and realistic technique, it is constrained by spatial limitations and often inefficient - as in reality. In contrast, continuous virtual movement without moving physically mostly resolves these constraints, but often leads to cybersickness. Therefore, the selection of a navigation technique for any use case is usually a trade-off decision. In this thesis, we therefore investigate and propose navigation support techniques to mitigate the negative aspects of these trade-offs. First, we examine ways to manipulate a user's movement to increase the share of real walking relative to the overall amount of navigation. In this context, we present a navigation technique that repositions the user (e.g., to the center of the available real-world space) while they travel to a target in order to maximize the interaction space and facilitate real walking at the chosen location. Furthermore, we investigate whether rotation gain (the manipulation of the mapping between physical and virtual head rotations), which can be used for techniques such as redirected walking, is usable in screen-based setups such as CAVE-like environments. Second, we propose navigation support techniques based on an automated analysis of the virtual scene that are aimed at making navigation more efficient. To this end, we analyze scene visibility (which parts of the scene are visible from any viewpoint) and the evaluation of the quality of such a viewpoint (how informative it is). First, we present a comparison of different measures to estimate viewpoint quality along with novel methods of our own. Furthermore, we introduce an approach for approximating scene visibility efficiently by restricting the computation to navigable areas of a scene. In addition, we show that these techniques can be used successfully to increase the efficiency of navigation, by integrating them into different navigation interfaces. In this context, we propose a technique to adjust travel speed automatically, reducing the amount of required cognitive resources, an approach to automatically choose intermediate target locations during long-distance travel that reduces cybersickness, and an exploration assistance technique that improves the user's cognitive map build up to allow for more efficient subsequent navigation.