This paper proposes a remote monitoring system on an embedded Linux platform for Global Positioning System (GPS) and vision hybrid positioning. As widely known, GPS signals are blocked by skyscrapers, trees, and multi-layered complex road structures in urban areas. Due to the limitations in satellite visibility, the positioning accuracy, availability, and continuity cannot be ensured in urban area. The proposed remote monitoring system integrates the off-the-shelf sensors on an embedded Linux platform for enhanced modularity and portability. The proposed system transfers the measurements of a GPS receiver and a vision sensor acquired by a small low-cost embedded Linux platform to the central monitoring station. The central host computer processes differential GPS measurements and image streams to generate robust hybrid positioning results. To mitigate errors occurring due to mismatched sampling time instants of the GPS receiver and the vision sensor, an efficient scheme is adopted for time synchronization. To verify the initial feasibility of the proposed system, an experiment was performed utilizing filed-collected measurements.

Keywords: hybrid positioning system, GPS, vision sensor, embedded Linux system