An Exploration of Techniques to Improve Relative Distance
Judgments within an Exocentric Display
6.1 Study Limitations
There are limitations within this study that should be noted. Most notably, the world image was positioned, via the software, to appear at approximately ten feet from the observer's eyepoint. This was done so that the world plane and the five cubes could be seen entirely within the display viewport at all times. Had the world been scaled to a smaller size and positioned closer to the observer, the features of head-motion tracking and stereo viewing may have proved to be more beneficial.
In the case of head-motion tracking, a closer image would have afforded a greater perspective change (rotation) when subjects moved their heads to look at the world. Another way to increase the world perspective change (arising from head-motion tracking) would have been to exaggerate the effect of head translation. It remains to be seen if increased head-tracking response would significantly alter the conclusions of this study (since subjects manually rotated the world at a minimum angle likely to be much greater than the gain provided by such techniques), but the increased head-tracking response may provide a more fair evaluation.
Similarly, studies have shown stereopsis to be most beneficial when objects of interest are within approximately one meter from the viewer (Johnston, Cumming and Parker, 1991); at over ten feet away, the world may have been perceived as too far away for stereo to be helpful. Again, had the world been scaled down and imaged closer to the viewer, stereo viewing's benefit for a relative distance task may have changed.
6.2 Design Recommendations
The research reported here, as part of a larger on-going project on communicating situation awareness, provides several potentially important contributions to the design of displays for non-immersive virtual environments. The findings indicate that in order to accurately perform a relative distance judgment task within an exocentric spatial display, a user should be given the ability to significantly change his or her perspective view of the 3D scene. Within the range of conditions tested here, we conclude that providing the ability to see in stereo, or tracking the user's head movements in order to appropriately change the world view, does not increase performance on a relative distance judgment task within an exocentric display. This is also true even when these features are provided in conjunction with an image rotation capability. Additionally, the provision of image rotation appears to increase user confidence in their distance judgments.
In an effort to determine an effective method for providing the different perspective views, it was observed both quantitatively and qualitatively that users should be given the ability to control the changes in the world view. However, providing a method by which users can quickly access orthogonal views of the world was also an important consideration, especially when performing time-critical judgments. Subjects found this rotation technique (known in the experiment as "Discrete Views"), in conjunction with Manual Rotation, to be most useful. This technique appeared to support a range of strategies which subjects employed in making relative distance judgments within the spatial display. While Manual Rotation seemed to be an effective technique for those subjects who used motion parallax as a cue to relative distances, Discrete Views was more extensively used by those who attended to the 2D distances in the various views to make their relative distance determination. By providing both techniques, greater flexibility in use and higher user satisfaction might be accomplished.
It was also shown that Discrete Views (even in the absence of Manual Rotation) reduced the time taken to make relatively accurate distance judgments. In the second experiment, it took approximately 32 seconds for a user to make an accurate relative distance judgment using a Manual Rotation technique and approximately 24 seconds to make a judgment using only the Discrete Views technique. As suggested in the third experiment, a manual rotation technique can be considered an effective tool for making a relative distance judgment, for applications which allow at least 24 seconds for a decision. For decisions that need to be made in less time, there is wide subject variability in the effectiveness of this technique. It should also be noted that the 24 seconds includes both the length of time to make a decision and the length of time required to manipulate the world, which involved using the keyboard in these experiments. Other world manipulation interfaces may decrease the amount of time taken to manipulate the viewpoint and may reduce the cognitive load introduced by the control interface.
6.3 Future Research
The results of this research suggest a number of hypotheses for future research. In particular, the results from the first experiment imply that stereoscopic viewing and head-motion tracking contribute little to relative distance judgments within exocentric displays. It would be important to see if these features become more significant when viewing more visually complex scenes or ones that contain objects which are imaged closer to the viewer. Also, the results reported here are for exocentric displays. Would the same techniques be of value in an egocentric display, and again, would stereoscopic viewing and head-motion tracking play a bigger role?
The manipulation of the world required the use of the keyboard. Determining the best interface for changing the world viewpoint (with respect to completing the task) would also be of interest. Independent variables could include a keyboard, joystick, wand or perhaps a representation of the world perceived as being held in one's hand. Such a representation would allow one to "pick up" the world and rotate it as one would pick up and handle any similarly sized object.
Throughout the experiments reported here, subjects detailed a number of different strategies that they used in order to make relative distance judgments within a given display condition. It would be interesting to see if a particular strategy that was highly effective for one individual could be taught and used equally well by another individual. If so, then this skill may be acquired for judgment tasks with shorter time constraints. In addition, it would be interesting to see how these experimental findings would apply to real world situations. Incorporation of this particular task, in conjunction with other tasks, may influence the application of these research results due to context effects.