NaviGlass: Indoor Localisation Using Smart Glasses

Smart glasses (e.g. Google Glass) is a class of wearable embedded devices with both inertial sensors and camera onboard. This paper proposes a smart glasses based indoor localisation method called NaviGlass. Because of high energy consumption of vision sensors, NaviGlass uses inertial sensors predominantly and uses the camera images for correcting the drift in the position estimates due to the accumulated errors of inertial sensors. On account of limited computation resources available on smart glasses, the computation time for image matching, which is needed to correct the position estimate, is high. We propose a feature reduction method that can significantly reduce the computation time for image matching but with little compromise on accuracy. We compare our method against Travi-Navi, which is a state-of-the-art localisation system that uses both inertial and image sensors. Our evaluations show that our proposed method achieved a mean localisation error of 3.3m which is 64% less than that of Travi-Navi.