Thumbnail Image

Human Visual Navigation: Effects of Visual Context, Navigation Mode, and Gender

Abstract This thesis extends research on human visual path integration using optic flow cues. In three experiments, a large-scale path-completion task was contextualised within highly-textured authentic virtual environments. Real-world navigational experience was further simulated, through the inclusion of a large roundabout on the route. Three semi-surrounding screens provided a wide field of view. Participants were able to perform the task, but directional estimates showed characteristic errors, which can be explained with a model of distance misperception on the outbound roads of the route. Display and route layout parameters had very strong effects on performance. Gender and navigation mode were also influential. Participants consistently underestimated the final turn angle when simulated self-motion was viewed passively, on large projection screens in a driving simulator. Error increased with increasing size of the internal angle, on route layouts based on equilateral or isosceles triangles. A compressed range of responses was found. Higher overall accuracy was observed when a display with smaller desktop computer monitors was used; especially when simulated self-motion was actively controlled with a steering wheel and foot pedals, rather than viewed passively. Patterns and levels of error depended on route layout, which included triangles with non-equivalent lengths of the two outbound roads. A powerful effect on performance was exerted by the length of the "approach segment" on the route: that is, the distance travelled on the first outbound road, combined with the distance travelled between the two outbound roads on the roundabout curve. The final turn angle was generally overestimated on routes with a long approach segment (those with a long first road and a 60° or 90° internal angle), and underestimated on routes with a short approach segment (those with a short first road or the 120° internal angle). Accuracy was higher for active participants on routes with longer approach segments and on 90° angle trials, and for passive participants on routes with shorter approach segments and on 120° angle trials. Active participants treated all internal angles as 90° angles. Participants performed with lower overall accuracy when optic flow information was disrupted, through the intermittent presentation of self-motion on the small-screen display, in a sequence of static snapshots of the route. Performance was particularly impaired on routes with a long approach segment, but quite accurate on those with a short approach segment. Consistent overestimation of the final angle was observed, and error decreased with increasing size of the internal angle. Participants treated all internal angles as 120° angles. The level of available visual information did not greatly affect estimates, in general. The degree of curvature on the roundabout mainly influenced estimates by female participants in the Passive condition. Compared with males, females performed less accurately in the driving simulator, and with reduced optic flow cues; but more accurately with the small-screen display on layouts with a short approach segment, and when they had active control of the self-motion. The virtual environments evoked a sense of presence, but this had no effect on task performance, in general. The environments could be used for training navigational skills where high precision is not required.
Type of thesis
Thomson, D. M. (2012). Human Visual Navigation: Effects of Visual Context, Navigation Mode, and Gender (Thesis, Doctor of Philosophy (PhD)). University of Waikato, Hamilton, New Zealand. Retrieved from https://hdl.handle.net/10289/6374
University of Waikato
All items in Research Commons are provided for private study and research purposes and are protected by copyright with all rights reserved unless otherwise indicated.