Abstract
Purpose: :
Depth information derived from stereo or using active sensors is robust to changes in lighting, contrast, and texture. Depth perception is critical to negotiating overhanging obstacles, which are an important hazard for the visually impaired. We evaluated depth information as a visual representation for mobility for normally-sighted volunteers using simulated prosthetic vision, including in the presence of overhanging obstacles.
Methods: :
4 normally-sighted participants (20/20, Peli-Robson>=0.95) used a mobile artificial vision stimulator set with 30x35 pixels centrally displayed identically to both eyes to navigate a mobility course. A shroud blocked their normal vision. Each phosphene corresponded to the depth of the corresponding visual field. Participants traversed a course viewing color images at full-resolution to establish their preferred walking speed. The course consisted of 6x3 1.5m cells with removable curtains at intersections. Paths were changed to avoid memory effects. Unfamiliar overhanging obstacles of varying size were introduced to some trials. Institutional Ethics Board approval was obtained.
Results: :
Depth is an effective representation for mobility: average percentage of preferred walking speed (PPWS) was significantly >50% in trials with (n=24) and without (n=30) obstacles (p<0.05). No significant difference was observed in PPWS when comparing trials with and without obstacles (p=0.24).
Conclusions: :
Participants were able to effectively navigating using depth as a visual representation, at greater than half their preferred walking speed. They showed little difficulty in adapting to the substitution of depth, and negotiated suspended obstacles without significant disruption. Depth information shows promise as a visual representation to be used on a retinal prosthesis. We are conducting further trials to compare depth with standard phosphene approaches.