Indoor Mobility(Full Phases of Indoor Navigation Services)

Over the last decade, navigation systems have been widely used as resources for path planning and way finding. Normally, they are tracked and guided by specific positioning technologies. Outdoor navigation services have been developed over the years by GPS (Global Position System), which is a quite common precise and infrastructure-free solution in large environments. However, the available navigation services in indoor spaces do not complete, and no standard has developed to fulfill all indoor navigation requirements. In this thesis I discuss about indoor navigation requirements, like this question “How can a human being find his/her destination in a building he / she does not know?”, etc. Here I propose an indoor navigation system that contains the functional modules and related techniques. The functional modules are Indoor Mapping, Indoor Positioning, Path Planning, En-route Assistant and Analysis respectively, which support the full phases of indoor navigation services. Indoor Mapping: A 2D and 3D indoor mapping approach is proposed, which 1) can build an effective 2D SVG map that updates selectively, 2) converts it into a 3D virtual scene automatically at back-end, 3) includes an information management system at back-end, 4) renders and interacts efficiently and effectively on a smart phone, and 5) offers a vivid virtual indoor visualization for virtual navigation services at front-end. Indoor Positioning: I propose two magnetic field positioning techniques: Robot Simulation-oriented Mobile Localization (RSML) and Computer Vision-oriented Fingerprint Localization (CVFL). 1) RSML focuses on a probabilistic framework that measures the relative and absolute position to simulate a robot movement. I propose an integrated magnetic field positioning approach based on eXtended Particle Filtering (XPF) algorithm, which fuses magnetic fingerprints, Wi-Fi fingerprints, and Pedestrian Dead Reckoning; 2) CVFL focuses on Magnetic Fingerprint Image-rization (MFI). It converts fingerprints to images and classifies them by CNN training. Indoor Path Planning: Wayfinding is necessary to reach the destination, which includes route planning and path navigation. An optimized ant colony algorithm is developed to avoid the obstacles (walls). En-route Assistant: En-route Assistant module provides the services in various devices, handed system, wearable system, etc. can be used for an enhanced navigation. In real-time navigation, it provides instructions based on a convenient trip for citizens and safe trip plan for VIPs (Visually Impaired People) and MIPs (Mobility Impaired People). Indoor Data Analysis: Here, a POI recommended algorithm is proposed based on social relations. It uses social relations to enhance the accuracy of the recommended algorithm from a large number of user behavior data in social network. Indoor Mobility also provides a social relationship mining model to recommend POI by a classification mapping between social relationship and POIs.