Bending the Curve: Sensitivity to Bending of Curved Paths and Application in Room-Scale VR

Abstract

Redirected walking (RDW) promises to allow near-natural walking in an infinitely large virtual environment (VE) by subtle manipulations of the virtual camera. Previous experiments analyzed the human sensitivity to RDW manipulations by focusing on the worst-case scenario, in which users walk perfectly straight ahead in the VE, whereas they are redirected on a circular path in the real world. The results showed that a physical radius of at least 22 meters is required for undetectable RDW. However, users do not always walk exactly straight in a VE. So far, it has not been investigated how much a physical path can be bent in situations in which users walk a virtual curved path instead of a straight one. Such curved walking paths can be often observed, for example, when users walk on virtual trails, through bent corridors, or when circling around obstacles. In such situations the question is not, whether or not the physical path can be bent, but how much the bending of the physical path may vary from the bending of the virtual path. In this article, we analyze this question and present redirection by means of bending gains that describe the discrepancy between the bending of curved paths in the real and virtual environment. Furthermore, we report the psychophysical experiments in which we analyzed the human sensitivity to these gains. The results reveal encouragingly wider detection thresholds than for straightforward walking. Based on our findings, we discuss the potential of curved walking and present a first approach to leverage bent paths in a way that can provide undetectable RDW manipulations even in room-scale VR.

A Review of Requirements and Approaches for Realistic Visual Perception in Virtual Reality

“…In this methods there is still a lack of balance information and therefore it can still cause cybersickness. Langbehn et al (2017) try to redirect the users movement to fit large environments into smaller tracking areas. Benefit of this method in comparison to the other ones, is it's multi-user potential and the presence of correct walking cues to our sense of balance.…”

On VR Spatial Query for Dual Entangled Worlds

“…Redirected Walking (RW) [43] introduces minor differences in the walking speed and turning angle to adapt the perception of walking in the dual worlds. Basic RW operations include Translation (TO) [59], Rotation (RO) [24,37], and Curvature (CO) [26,35]. TO introduces a slight scaling factor between the walking speed in the virtual world and the actual walking speed in the physical world.…”

“…Compared with DP, DEWN generates feasible solutions with much smaller time because the pruning strategies effectively trim off redundant 25 https://bit.ly/2DZXQLv, https://bit.ly/2DZXS65. 26 The estimated RW cost for a v-edge with length l is set to β (l ) (the MIL upper bound) as a safe estimate, since COLA does not process the physical world. 27 V-path lengths are averaged only from feasible solutions.…”

Object Manipulation in Virtual Reality Under Increasing Levels of Translational Gain

“…The seminal redirected walking research (e.g., [42][43][44]) mapped physically curved paths to virtual straight paths to test the illusion, but more recent work has looked at the tolerance for detection of curved physical paths, either simply walking through a room [28], or in conjunction with touching a wall surface [32,33]. The research suggests that there are larger tolerances (i.e., lowered sensitivity) to variations in the curvature of physical vs. virtual paths, compared to straight physical vs. curved virtual paths, and curved walls can be perceived as straight walls [32,33].…”

“…Translational gain has frequently been overlooked in favour of both redirected walking and locomotion interfaces. Redirected walking uses combinations of translational [21], rotational [27] and curvature [28] gain alongside other perceptual tricks to portray infinitely traversable virtual spaces. However, these techniques have significant limitations regarding physical requirements (minimum 6m 2 [45] but realistically up room-scale to 22m 2 [19]), and they cannot support singleroom virtual spaces where every visible corner is consistently reachable and fully explorable.…”