Accurate and reliable real-time indoor positioning on commercial smartphones

摘要

This paper outlines the software navigation engine that was developed by SPIRIT Navigation for indoor positioning on commercial smartphones. A distinctive feature of our approach is concurrent use of multiple technologies for indoor positioning. Measurements from such smartphone sensors as IMU (3D accelerometer, gyroscope), a magnetic field sensor (magnetometer), WiFi and BLE modules, together with the floor premises plan are used for hybrid indoor positioning in the navigation engine. Indoor navigation software uses such technologies as PDR, Wi-Fi fingerprinting, geomagnetic fingerprinting, and map matching. Being blended in the particle filter, dissimilar measurements allow solving a set of principal tasks. First, the navigation engine can automatically start in any place of a building wherever user switches on his or her smartphone. There is no need to enter initial position manually or to start outdoors where initial position can be determined by GPS/GNSS receiver. Then, operating in the tracking mode, the navigation engine provides real-time indoor navigation for displaying current user position either on the floor plan or on Google Indoor Map if the latter is available for the building. At last, the navigation engine can recover tracking from failures that are the known problem of the particle filter occurring when all particles are accidentally discarded. The automatic recovery of tracking in this case allows continuing tracking and increasing availability of indoor navigation. The navigation engine exits in a form of SDK that serves for building mobile applications both for Android and iOS. Positioning results given for different indoor environments in a shopping mall and in a big exhibition hall show fast TTFF indoors and accurate and reliable real-time indoor positioning with accuracy of about 1–2 m.