Towards Enabling More Effective Locomotion in VR Using a Wheelchair-based Motion Platform

Abstract

This paper addresses two questions relevant to the design of effective locomotion methods for VR using a novel wheelchair motion simulation interface. First, we investigate the extent to which people's ability to keep track of where they are in an immersive virtual environment can be facilitated by actual physical movement (rotation and translation) in the context of vehicular travel. Second, we quantitatively analyze various characteristics of the travel paths produced by different types of locomotion control systems to gain insight into the aspects of control that can evoke or impede natural patterns of movement through a virtual environment. In a within-subjects experiment, we asked 35 volunteers to virtually search through 16 identical-looking boxes randomly placed within a realistically rendered, circularly symmetric virtual room to find 8 hidden targets. Participants performed this task under four different conditions of integrated visual and physical movement, controlled via a joystick interface attached to a motorized wheelchair. In all four cases participants 'drove' their virtual viewpoint using the joystick, but the nature of the accompanying physical movement varied between the conditions. The four conditions were: no physical movement; full physical rotation only; full physical translation and rotation; and ''partial'' physical translation and rotation, wherein the extent of the actual physical movement was proportionally reduced relative to the visually-indicated movement. Analysis of the search results did not find a statistically significant main effect of the physical movement condition on total distance traveled or total number of revisits to previously searched locations. However we did see a trend towards greater search accuracy in the full physical motion condition, with a greater proportion of perfect trials, a smaller proportion of failed searches, fewer boxes revisited on average, and more novel boxes searched before the first revisit in that condition than in the others. Analyzing the paths traveled, we found that the velocity and curvature profiles of the virtual motion paths enabled by our novel joystick-controlled wheelchair motion simulation interface were more qualitatively similar to those produced by natural walking than were travel paths we had previously observed when more basic joystick locomotion control methods were used. This suggests potential benefits in adopting a vehicle-simulation movement control method for joystick locomotion control in VR.