Abstract
Purpose: Factors expected to improve the immersive experience (‘realism’) in a virtual environment (e.g. updating the viewpoint using head position) may affect collision detection performance.
Methods: Participants stood or walked on a treadmill 75cm from 95-degree-wide screen that displayed a textured representation of a ‘shopping mall’ corridor. Participants indicated whether they would collide with human-sized obstacles, that appeared for one second, if they continued on the same path. Perceived safe passing distance and decision quality were compared for: (1) locomotion (standing versus walking at fixed and participant-controlled speeds); (2) viewing eyes (monocular or binocular); (3) viewpoint updating (moving the viewpoint with head position) and (4) viewpoint error (30cm left or right).
Results: When standing, participants had a slightly smaller perceived safe passing distance (p=0.07) and more definite decisions (p=0.01) than when walking. Participant-controlled walking speed, viewing eyes and viewpoint updating had no significant effect on the perceived safe passing distance (p>0.18) or decision quality (p>0.38). Viewpoint error caused subjects to have a non-symmetric perceived safe passing distance (shifted their perceived centerline), with females shifting towards and males shifting away from the incorrect viewpoint (p=0.02).
Conclusions: Attempts to increase realism (walking, viewpoint updating, using one eye with non-stereo display) did not affect collision detection performance. Temporal lag of our headtracker may have degraded performance in the viewpoint-updating condition. The implications of the gender difference in perceived body position due to viewpoint error is not clear.
Supported by NIH grant EY12890.