A wearable indoor navigation system for blind and visually impaired individuals.

摘要

Indoor positioning and navigation for blind and visually impaired individuals has become an active field of research. The development of a reliable positioning and navigational system will reduce the suffering of the people with visual disabilities, help them live more independently, and promote their employment opportunities.;In this work, a coarse-to-fine multi-resolution model is proposed for indoor navigation in hallway environments based on the use of a wearable computer called the eButton. This self-constructed device contains multiple sensors which are used for indoor positioning and localization in three layers of resolution: a global positioning system (GPS) layer for building identification; a Wi-Fi - barometer layer for rough position localization; and a digital camera - motion sensor layer for precise localization. In this multi-resolution model, a new theoretical framework is developed which uses the change of atmospheric pressure to determine the floor number in a multistory building. The digital camera and motion sensors within the eButton acquire both pictorial and motion data as a person with a normal vision walks along a hallway to establish a database. Precise indoor positioning and localization information is provided to the visually impaired individual based on a Kalman filter fusion algorithm and an automatic matching algorithm between the acquired images and those in the pre-established database. Motion calculation is based on the data from motion sensors is used to refine the localization result. Experiments were conducted to evaluate the performance of the algorithms. Our results show that the new device and algorithms can precisely determine the floor level and indoor location along hallways in multistory buildings, providing a powerful and unobtrusive navigational tool for blind and visually impaired individuals.