|
|
Kevin Arthur
Department of Computer Science
University of North Carolina
Chapel Hill, NC 27599-3175 USA
+1 919 962 1729
E-mail: http://www.cs.unc.edu/~arthur/
© Copyright on this material is held by the author.
The goal of this research is to quantify the effects of a head-mounted display's field of view (FOV) on human performance of 3D tasks representative of those typically performed in virtual environments.
Keywords
Head-mounted display, field of view, task performance, adaptation, spatial awareness, presence, simulator sickness.
Head-mounted displays are tools that allow us to perform tasks in 3D virtual environments, borrowing from our natural abilities to perform 3D tasks in the real world. An important parameter in choosing or designing an HMD is the field of view (FOV) of the display.
In the real world, our effective FOV spans approximately 200 degrees horizontally by 150 degrees vertically. Many commercially available HMD's have relatively narrow fields of view, ranging from roughly 30 to 70 degrees diagonally.
Narrow FOV has been shown (in real environments) to degrade human performance on navigation, manipulation, spatial awareness, and visual search tasks, and to disrupt our eye- and head-movement coordination and our perception of size and space [1][2].
Wide FOV displays are not yet generally available, and even when the engineering difficulties of realizing them are overcome, choosing the widest FOV possible may not be optimal for many applications. A wide FOV will aggravate simulator sickness effects, and in particular those due to vection and visual-vestibular mismatch [4] [9], and wide FOV may be unnecessary for tasks that are localized within a small spatial region of interest.
This research will address the following research questions through controlled user studies on a set of tasks that is representative of many of those typically performed in virtual environments applications.
The remainder of this document discusses the four classes of tasks that are being studied under different FOV conditions. This is followed by a discussion of simulator sickness factors related to FOV.
Peripheral vision is known to be well-suited to maintaining self-orientation during locomotion[5], and there is evidence to suggest that a narrow field of view detracts from a person's ability to navigate through an environment effectively. Alfano and Michel [1] measured the ability of subjects to walk along a path while wearing goggles to restrict their FOV and found that a FOV of 12 or 40 degrees resulted in a significant errors and less pronounced effects were still present with a FOV of 90 degrees.
The present research will include studies to measure a subject's ability to walk through a complex virtual environment while avoiding collisions with obstacles. The task will be duplicated in a ``real world'' condition, in which the subjects will wear a mock HMD having weight, center of gravity, and FOV equivalent to one of the real HMD viewing conditions.
The hypothesis is that a larger FOV will cause fewer object collisions because subjects will be able to see more of the environment and also because more peripheral vision will give them a better sense of their movement and of their own position and orientation in the environment.
Tasks involving manual manipulation using one or both hands may be considered as having two phases: reaching (rapidly moving the hand to the vicinity of a target for manipulation) and grasping (using the fingers and thumb to have an effect on the object). There is evidence to show that grasping ability is not significantly affected by FOV, but that reaching ability is degraded by narrow FOV. This manifests itself in misjudged distances to objects [7].
A reaching task will be tested under different FOV conditions. The hypothesis for this task is that response time will vary with the distance to the target and that response time will likely have two components: the time to locate the object visually (by moving the head) and the time to reach for the object. Another hypothesis is that subjects may misjudge the position of the targets in a direction to or from their own position depending on FOV, and that subjects' consistency in judging distances may vary with FOV.
For virtual environment applications such as architectural walkthrough, it is vital that the participant obtain an accurate spatial layout and size of objects in a scene. Researchers have reported that a narrow FOV tends to make objects appear nearer than they really are and to effectively shrink the environment around the user [1][2].
In the present work, this will be tested in conjunction with the navigation study. After subjects perform a navigation task, they will be removed from the virtual environment and given a spatial memory task. They will be asked to report the presence or absence of various objects that were in the environment and the spatial arrangement of the objects.
The hypothesis is that with narrow fields of view subjects may miss some of the objects or misjudge their position within the environment, and may underestimate the size of the environment as a whole.
To some extent, most everything to be done in a virtual environment involves visual search. To navigate we need to find the place we wish to move to; to reach for something we usually need to locate it visually; to ``take in'' an environment and form a mental model of it we need to scan it visually. Visual search is a task that has been widely studied [8]. Peripheral vision is important for searching because, while we do not use the periphery to recognize and identify objects, visual events in the periphery serve to trigger changes in gaze [5].
A study is being conducted to measure the effect of FOV on a search task that involves finding and identifying randomly placed targets. This task will measure search time as a function of FOV and measure the magnitude and frequency of head movements as a function of FOV.
Wide FOV has been shown to cause increased simulator sickness effects due to vection [9]. Vection refers to the illusion of self-motion (we perceive ourselves to be moving when our body is not), and it has been cited as a key indicator of simulator sickness [4] [6].
To assess the effect of FOV on simulator sickness, the present studies will include the administration of a standard simulator sickness questionnaire [3].
1. Patricia L. Alfano and George F. Michel. Restricting the field of view: perceptual and performance effects. Perceptual and Motor Skills, 70(1):35-45, 1990.
2. Hubert Dolezal. Living in a world transformed: perceptual and performatory adaptation to visual distortion. Academic Press, 1982.
3. R.S. Kennedy, N.E. Lane, K.S. Berbaum, and M.G. Lilienthal. A simulator sickness questionnaire (ssq): A new method for quantifying simulator sickness. International Journal of Aviation Psychology, 3(3):203-220, 1993.
4. Eugenia M. Kolasinski. Simulator sickness in virtual environments. U.S. Army Research Institute Technical Report 1027, May 1995.
5. H.W. Leibowitz. Recent advances in our understanding of peripheral vision and some implications. Proceedings of the 30th Annual Meeting of the Human Factors Society, pages 605-607, 1986.
6. Michael E. McCauley and Thomas J. Sharkey. Cybersickness: Perception of self-motion in virtual environments. Presence, 1(3):311-318, Summer 1992.
7. Barbara Sivak and Christine L. MacKenzie. The contributions of peripheral vision and central vision to prehension. In L. Proteau and D. Elliot, editors, Vision and Motor Control, pages 233-259. Elsevier Science Publishers, 1992.
8. L.W. Stark, K. Ezumi, T. Nguyen, R. Paul, G. Tharp, and H.I. Yamashita. Visual search in virtual environments. Proceedings of the SPIE: Human Vision, Visual Processing, and Digital Display III, 1666:577-589, 1992.
9. Robert M. Stern, Senqi Hu, Richard B. Anderson, Herschel W. Leibowitz, and Kenneth L. Koch. The effects of fixation and restricted visual field on vection-induced motion sickness. Aviation, Space, and Environmental Medicine, 61(8):712-715, August 1990.