Abstract:
Effective indoor navigation is a challenging task for people with low vision, even they can potentially benefit from visual cue. Many researchers have proposed various approaches to improve the efficiency and accuracy of the navigation, for example, using audio cue as an assist. However, the visual rating has been neglected, which is important to provide high-quality assistant for different visual impairments. In this paper, we propose an augmented reality (AR) application that measures the visual rating. According to the visual rating, different navigation services, including visual aid and audio cue, are provided to users. Besides, object detection and depth estimation are adopted for detecting and avoiding obstacles, which can significantly improve the performance of the navigation. For the experiment, the Snellen chart is displayed on Microsoft Hololens2 and a pilot study has been conducted to verify its feasibility. The guidance information are shown in various sizes, combining with audio cue, for different vision-rating users. The strategy on both visual aid and audio cue will be tested by a user study for the next step. The results of object detection and depth estimation have shown high accuracy on Microsoft Hololens2 and will be integrated into the application in the future.
