Gaze behavior during navigation with reduced acuity

Abstract

Navigating unfamiliar indoor spaces while visually searching for objects of interest is a challenge faced by people with visual impairment. We asked how restricting visual acuity of normally sighted subjects would affect visual search and navigation in a real world environment, and how their performance would compare to subjects with naturally occurring low vision. Two experiments were conducted. In the first, 8 normally sighted subjects walked along an indoor path, looking for objects placed at unpredictable intervals to the left and right of the path, and identified single letters posted on the objects. A head-mounted eye tracker was used to assess their gaze direction in the environment. For half the trials, blur foils were used to restrict visual acuity to approximately logMAR 1.65. Gaze behavior, travel time, and letter recognition accuracy were compared between blurred and unrestricted conditions. In the second experiment, the same procedure was conducted, but performance was compared between acuity-restricted normally-sighted subjects and subjects with naturally occurring low vision (mean acuity 1.09 logMAR, range 0.48-1.85 logMAR). In Experiment 1, neither Blur nor the Letter Recognition Task individually had a statistically significant effect on travel time. However, when combined, there was an interaction between the two that increased travel time by approximately 63%, relative to baseline trials. Blur modified gaze behavior such that subjects spent more time looking down toward the floor while walking, at the expense of time spent looking in other directions. During Letter Recognition Task trials with Blur, subjects spent extra time examining objects, though more objects were missed altogether. In Experiment 2, low-vision subjects spent more time looking toward the boundary between the floor and the wall, but gaze patterns were otherwise similar to acuity-restricted subjects with normal vision. Low-vision subjects were also more likely to miss objects compared to acuity-restricted subjects. We conclude that under conditions of artificially restricted acuity, normally sighted subjects look downward toward the floor more frequently while navigating and take extra time to examine objects of interest, but are less likely to detect them. Low-vision subjects tend to direct their gaze toward the boundary between the wall and the floor, which may serve as a high contrast cue for navigation.

Keywords: Acuity impairment; Gaze behavior; Low vision; Mobility; Navigation; Visual accessibility; Visual search.

Fig. 1.

A blueprint of the experimental space used for both experiments. The box labeled “START” is where subjects began each trial, while the green arrows indicate the path they walked around the room. The blue lines represent the chain boundary, which subjects held onto as they walked. The black circles indicate the locations of stanchions, which were used to hold up the chain. The red box on the left side represents a raised platform that objects were placed on, and which constituted the left side of the path for that section of the course. Each square on the background grid is scaled to one square foot.

Fig. 2.

The Tobii glasses, a head mounted eye tracker, with the forward facing video camera visible on the right arm. The monocular eye tracker is partially visible behind the forward facing camera. The bottom panel shows the glasses with acuity restricting blur foils attached.

Fig. 3.

A picture of the course layout showing the objects set up for the Letter Recognition Task with Artificial Puddles. In the Letter Recognition Task condition, objects had one letter pinned on them, and could be placed either on the left or on the right side of the path. During trials without the Letter Recognition Task, objects were lined up against the wall on the left side of this photo, with no letters attached to them.

Fig. 4.

Directional gaze categories. Panel A illustrates the categories along the horizontal axis (blue=left, pink=path inspection, orange=right), while panel B illustrates categories along the vertical axis (blue=above floor-wall boundary, pink=floor-wall boundary, orange=floor).

Fig. 5.

Effect of Trial Condition on Travel Time. Mean travel durations for normally sighted subjects under different trial conditions. Error bars represent 95% confidence intervals. Trial conditions are grouped by those that did not include Artificial Puddles (50% of all trials) on the left, and those that Included the Artificial Puddles on the Right. The orange bars refer to those of trials which included the Letter Recognition Task (50% of all trials). Travel time for each condition is specified at the base of each bar.

Fig. 6.

Mean travel times for normally sighted subjects with Artificial Blur (the controls for this experiment) and low vision subjects under different trial conditions. Error bars represent 95% confidence intervals. Trial conditions are grouped by those that did not include Artificial Puddles (50% of all trials) on the left, and those that Included the Artificial Puddles on the Right. The orange bars refer to those of trials which included the Letter Recognition Task (50% of all trials). Travel time for each condition is specified at the base of each bar.