Redirected Walking Research

With traditional immersive VE systems, the scale of a virtual simulation is typically limited to the size of the tracking area. Redirected Walking (RDW), a technique first examined by Razzaque, overcomes the physical limitations of a tracking area by imperceptibly and judiciously rotating the virtual scene about the user. These rotations systematically guide the user on a curved path within the real world while the user perceives that they are following a straight line in the virtual world. Appropriate application of these injected rotations can gradually direct users away from real-world obstacles such as walls or other users. If the rate of rotation introduced by RDW is small enough, it is not perceived by the user and he or she will subconsciously respond by turning in the opposite direction in the real world in order to maintain the desired heading in the VE.

Under RDW, the virtual position is likely to have little correspondence to any real-world position. This would be problematic in augmented reality or mixed-reality applications, in which the real and virtual worlds need to be aligned. However, it is perfectly acceptable in a traditional VE, in which the fully synthetic world need not correspond to any physical landmarks.

Generalized Redirected Walking

Generalized RDW algorithms operate with no knowledge of the virtual world in which the user is navigating and/or the task in which the user is engaged. Steering is performed on the basis of the physical position of the user, a velocity expressing their linear motion, and the angular velocity of their head movement. Generalized RDW algorithms attempt to keep the user as far from tracking area boundaries as possible by steering the user in a preset pattern or between a fixed number of steering points. Algorithms proposed in the literature include the "steer-to-center," "steer-to-orbit," and "steer-to-multiple targets."

HIVE researchers implemented a real-time generalized RDW algorithm and demonstrated its use for the first time. This work included experiments in which users navigated through a virtual forest of unlimited size while searching for items placed in the environment. During the course of these experiments, users typically traveled virtual distances which would have placed them well outside the physical tracking area. In spatial memory tasks conducted at the end of each trial, user performance was statistically equivalent under RDW and normal conditions, indicating that the use of redirected walking did not adversely affect their mental model of the environment. Surveys conducted at the end of each trial showed that the use of RDW did not lead to an increase in simulator sickness.

Fully Optimized Redirected Walking for Constrained Environments

Generalized redirected walking techniques work well when applied to wide-open virtual environments that provide little information about what actions the user might take in the future. Constrained environments, such as a grocery store with aisles and other building interiors, limit the user’s future actions to a small set of predictable user actions. By preprocessing the map of the virtual environment, the major corridors of travel can be identified and these corridors can be used to produce real-time probabilistic predictions concerning the user’s future actions. This, in turn, can allow RDW to utilize otherwise undesirable space near the physical limits of the tracking space during periods in which it is known that users will not spontaneously turn in that direction and make optimal choices regarding what type of steering to apply.

Multi-User Redirected Walking

One of the most significant shortcomings of HMD-based immersive virtual environments such as the HIVE is the lack of a capability to support multiple users simultaneously. It is advantageous to allow multiple users to simultaneously use the system and share the tracking area. However, immersed users are essentially blindfolded by the HMD and are walking freely in a limited physical space. Thus, collisions with other users would be inevitable. The HIVE researchers have begun work to extend generalized RDW concepts to prevent collisions with other users as well as tracking area boundaries. The techniques currently under investigation include predicting the users' paths and determining if the paths are likely to intersect. If so, RDW steering is applied in order to avoid the probable collision.