Generating indoor maps by crowdsourcing positioning data from smartphones

Indoor maps are highly essential for indoor positioning and location-based services. Applications providing navigation support to users are rendered useless without a map of the vicinity being available. Presently, floorplans of public locations are collected and maintained by designated organizations using methods that require excessive manual intervention. This process of creating a database of indoor maps is neither efficient nor scalable to the practically infinite number of public indoor places around the world. In this paper, we present a crowdsourcing algorithm to automatically create floorplans of buildings with zero prior information. The algorithm leverages the positioning data shared by pedestrians using smartphone-based navigation systems in the building. It expects only position fixes and associated uncertainties from the navigation systems and does not depend on any particular navigation algorithm. The available positioning data in a completely unknown building is essentially PDR-based and is known to be prone to high amounts of accumulated error primarily due to the lack of reliable error resetting techniques. The presented algorithm takes into account the possibility of such highly erroneous motion traces of pedestrians while trying to generate map as accurately as possible. As an added merit, the algorithm does not depend on the availability of Wi-Fi access points.