Multi-sensor multi-network seamless positioning with visual aiding

This paper presents a measurement fusion approach for seamless outdoor/indoor positioning utilizing multiple sensor and network measurements in addition to computer-vision based aiding. A Kalman filter is implemented for fusing WLAN, Bluetooth, and high-sensitivity GPS positioning results with accelerometer based motion information as well as digital compass and camera derived direction. The fusion filter provides improved positioning accuracy to a single-technology system and offers higher positioning availability in time. In the experiment conducted inside the third floor of a typical office building, the proposed fused solution achieves horizontal accuracy of around 7 m with a 1-Hz update rate. Aiding the fusion filter with a heading rate measurement derived from visual processing improves the horizontal position accuracy further to around 6 m.