Towards a robust smartphone-based 2,5D pedestrian localization

Automatic real-time localization of people inside buildings is a huge challenge. For demanding applications in building services, different sensors (e.g. WLAN, RFID, UWB, or ultrasound) are currently used for the real-time indoor positioning. Most systems are only suitable for specific applications or are used under certain conditions, e.g. additional infrastructures and/or sensory mechanisms are needed. Smartphones, as widespread low-cost multi sensor systems, seem to be a promising platform for mass-market indoor localization applications. In this paper we present an approach for a smartphone-based pedestrian positioning inside of buildings. The novelty of this approach refers to further investigations towards a robust smartphone-based 2.5D pedestrian localization inside of buildings. Therefore, comprehensive analyses concerning barometric height estimations were carried out to expand the predominant 2D position into the third dimension. The 2D position is estimated according to the principle of dead reckoning. Furthermore an extended algorithm based on Sequential Monte Carlo methods for sensor fusion is presented which also integrates building models and a magnetic indoor positioning system.