Purchase this article with an account.
Stephen Lloyd Hicks, Iain Wilson, Joram Jacob van Rheede, Robert E MacLaren, Susan M Downes, Christopher Kennard; Improved mobility with depth-based residual vision glasses. Invest. Ophthalmol. Vis. Sci. 2014;55(13):2153.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The majority of legally blind individuals retain some residual vision. While this may be insufficient for obstacle detection it is often enough to perceive a bright nearby display. We developed a wearable visual aid that displays the size and distance of nearby objects as bright regions of light. We reasoned that such a display could provide an intuitive and simplified field of view that would permit the wearer to detect and avoid obstacles at a distance.
Residual Vision Glasses (RVG) were built from two small OLED displays mounted within 3 cm of the eye, without focussing optics. A head-mounted depth camera (Primesense) provided real-time three dimensional scene data which was converted into a depth map and presented on the displays such that nearby objects appeared brighter (Figure 1). 4 controls and 12 sight impaired participants with conditions including AMD, retinitis pigmentosa and retinal dystrophy were recruited from the local population; average (± 1SD) age = 54yr (± 18) LogMAR = 1.26 (± 0.3). Each participant was given a short orientation with the RVGs and then asked to navigate an obstacle course 15 x 5 metres consisting of a set of 1.2 m foam pillars or a set of doorways. Participants performed both tasks with and without the RVGs. The depth detection range of the RVGs was artificially limited to 4 metres.
All participants were able use the RVGs to correctly report when they faced a nearby obstacle and the majority of participants could use the displays to walk the obstacle course. Performance was quantified as the distance at which a person began deviating away from a collision course with an obstacle. Unaided controls began deviating up to 5 metres, whereas unaided sight impaired participants began deviating between 1 and 4 metres from obstacles. When using the RVGs both controls and sight impaired individuals began deviating at an average distance of 3 metres. An improvement in object detection distance was found in the majority of participants (Figure 2).
With a small amount of training, sight impaired individuals were able to use RVGs to identify obstacles up to the distance limit set by the camera. In many cases this was an improvement over their unaided state. Further work is under way to assess the benefit of see-through displays and increased distance detection.
This PDF is available to Subscribers Only