GoIn—An accurate 3D indoor navigation framework based on light landmarks

Wireless computer network (WLAN)-based building-level positioning coupled with cellular information is thoroughly investigated in literature. This paper introduces a framework for accurate indoor positioning system (IPS) and navigation, which outperforms known algorithms. The main idea is to use a modified particle filter with weights computed as a function of radio frequency (RF) finger-printing, odometry, visual landmarks, and map constraints. Visual landmarks are extracted using a low resolution camera to track dominant landmarks such as lights. We demonstrate the efficiency of this framework on an Android-based mobile device on various indoor scenes with some prior knowledge. Our method allows a robust positioning compared with other systems evaluated under the same conditions, at 30 Hz and with a fairly low energy consumption.