A Mobile Indoor Localization System With Geomagnetic Dominant Characteristics

As the coming of mobile Internet age,all kinds of Location Based Services(LBS)depend on accurate and real-time locations.Especially in some specific indoor scenes,LBS is quite important,such as the maintenance of auto warehouse,the elderly care in headhouses and the passenger navigation in airports.Achieving an accurate real-time location can make people's life much more convenient and improve our work efficiency greatly.It has wide application prospects and big commercial value.In indoor environments,Global Position System(GPS)can hardly work due to the signal attenuation.Thus,there is an urgent need for a general solution for common scenes.Since twenty-first Century,plenty of researches have extensively explored the software and hardware solution scheme of indoor localization.But up to now,there isn't a mature solution that reaches a satisfactory balance among the accuracy,the usability and the installation maintenance.Our work takes the smartphones as a primary platform.As is well known,currently smartphones have very high coverage,powerful processors and get equipped with a variety of sensors,which is the most ubiquitous and most easy-to-use terminal in mobile Internet age.It means smartphones could have the opportunity to traverse most of indoor environments and provide tremendous amount of environment information which is crucial for indoor map construction or localization.Such a characteristic makes smartphones perfectly suitable for indoor localization.This paper takes a deep investigation into the domestic and foreign research status of indoor localization.Based on the existing problems including high energy consumption,demanding extra infrastructures,etc.,we choose the geomagnetic location method and introduce 2-D geomagnetic fingerprints.A completely geomagnetic dominant indoor localization and navigation system is designed and implemented.This system takes the build-in magnetometer of smartphones as the dominant sensor,with the initial inertial measurement unit(IMU)and the inclinometer supplemented.Based on the characteristic of 2-D fingerprints,a matching scheme combining point features and line features is presented.The data collection and map construction is realized in a crowdsourcing framework.The system contains three modules including the data collection,cloud server and the localization and navigation.Than to reduce the difficulties in environment information collection,we put forward the scheme to exploring unknown areas automatically with an unmanned aerial vehicle based on visual SLAM.A dense map construction thread and an exploring thread are added to the ORB-SLAM system to realize accurate and efficient map construction.Our contributions can be summarized as follow: 1)2-D geomagnetic fingerprints with rich features are introduced,which improve the discrimination and reliability greatly compared to classic 1-D fingerprints;2)Based on 2-D fingerprints,we explore the possibility of pointto-point matching in geomagnetic localization,and realize a complete geomagnetic dominant indoor localization system,which reduces energy consumption greatly and needs no extra infrastructures;3)we design a simple human-computer interface to achieve accurate 2-D point fingerprints,which obtains the necessary data at a very low cost;4)we design a set of corresponding data collection,map generation and data fusion algorithms and introduce overall data preprocessing and optimization to improve the performance;6)we implement our system in Android smartphones and conduct a series of experiments in an office building to evaluate the system comprehensively;5)we design an automatic geomagnetic fingerprint map construction scheme based on visual SLAM and unknown environment path planning,which provide a new way to achieve accurate fingerprint maps.