부산대학교 도서관

Virtual reality (VR) systems allow a user to explore virtual environments (VEs) intuitively by linking display and interaction to physical movements in the real world. Walking is an intuitive method of traversal, but challenging for VR; VEs do not need to match physical spaces, so users may encounter obstacles in the real world. Redirected walking (RDW) is a technique which remaps a user's physical walk onto a subtly different virtual path. The user then unknowingly adjusts their physical path to account for the change. With carefully selected transformations the user can be steered away from physical obstacles, allowing free walking in the VE. However, state of the art RDW techniques still require a large amount of physical space. The work in this thesis aims to reduce physical space requirements for RDW techniques. Certain RDW tasks such as infinite straight-line walking require large amounts of physical space due to perceptual limits. However, VEs which contain obstacles may not contain long straight paths and can be analysed to provide useful information about future user walk directions. This research therefore focuses specifically on the application of RDW in obstacle-rich VEs to small physical spaces. We present the following contributions on this theme: (1) MCRDW, a gain selection algorithm for RDW which uses simulated walks to anticipate future user trajectories, (2) a psychophysical study on tolerance to rate of gain change, the results of which indicate that slow gain change is significantly harder to detect than sudden gain change, and (3) Shared Spaces, a multi-user technique to allow users to share spaces virtually while allowing real walking locally.