PhD Proposal: Methods for Natural Walking in Virtual Reality
Locomotion is a fundamental component of experiences in virtual reality (VR). However, locomotion in VR is often difficult because the layouts of the physical and virtual environments are often different, which may cause unobstructed paths in the virtual world to correspond to obstructed paths in the physical world. Thus, in order to deliver a comfortable and immersive virtual experience to users, it is important that the user can explore the virtual world using techniques that help them avoid collisions with unseen physical objects. Redirected walking (RDW) is one such technique that enables collision-free locomotion in VR using real walking. Prior research has shown that RDW has promise as a natural, easy-to-use locomotion interface for VR, but two major constraints still limit its effectiveness and practicality outside of controlled lab environments. First, current RDW algorithms do not work well when the size and layout of the physical environment are significantly different from that of the virtual environment. This is because as the size and shape of the physical environment and virtual environment differ more, the likelihood that a valid path in the virtual environment corresponds to an invalid (collision-yielding) path in the physical environment increases. This problem is exacerbated for small physical environments where free locomotion is difficult (e.g. a living room). The second condition that is challenging for RDW is when the layout of the physical environment changes dynamically or is not fully known at runtime. If the user's physical surroundings are constantly changing or the positions of physical obstacles are unknown, it is difficult for the redirection algorithm to reason about the user's surroundings and steer them on a collision-free path.This thesis focuses on making progress towards creating a RDW system that is usable outside of controlled lab environments by mitigating the restrictions imposed by these two constraints. We achieve this using multiple techniques. First, we build upon the concept of alignment to develop new RDW algorithms that take into account both the physical and virtual environment to optimally steer the user away from physical obstacles. Additionally, we introduce new RDW algorithms that are built using a new formalization of the RDW problem based on motion planning. This mathematical formalization allows us to leverage techniques from robot motion planning and computational geometry to develop steering algorithms that are more easily generalizable to different environment layouts without requiring us to update the algorithm implementation. To further improve the usability of our RDW algorithms in dynamic, unknown environments, we propose to augment the traditional RDW pipeline with data about the user's physical environment that is computed in real time using computer vision. Finally, we also plan to improve the usability of RDW for a wide range of users in a range of VR system and virtual environment configurations by developing new methods to accurately estimate how much redirection can be applied before users begin to feel symptoms of simulator sickness.
Dr. Dinesh Manocha
Dr. Huaishu Peng
Dr. Aniket Bera
Dr. Evan Suma Rosenberg (University of Minnesota, Twin Cities)