A Pedestrian Dead Reckoning (PDR) Based Indoor Navigation System

With the rapid development of IoT(Internet of Things)application,and the huge marketing of smart city,the requirement of investigating and developing the low-cost,easy-to-deploy and easy-to-use indoor localization technologies is growing day by day.The popular application is indoor localization basd indoor navigation.This thesis proposes a indoor navigation system based on PDR which includes three modules:map construction,PDR,map matching and navigation.Specifically,this thesis focuses on the accuracy improvement of PDR system and real time navigation for users.The primary barriers for the adoption of indoor navigation include the lack of fine-grained and up-to-date indoor maps.This thesis investigates the feasibility of 3D-Model-based Indoor Mapping and applies SfM techniques to build 3D point clouds from unordered 2D photos and proposes extracting 2D floor plan from the 3D point cloud.This method make maps more accurate and has lower cost.Compared with WiFi fingerprint,pedestrian dead reckoning(PDR)is less affected by the complex indoor environment and has low demand for datebase and hardware equipments.PDR commonly use inertial sensors such as accelerometer,gyroscope and magnetometer to positioning.Typically,there are three phase in PDR: step detection,stride length estimation and heading determination.This thesis proposes using dynamic threshold to improve accuracy and sability.To compensate for accuracy of PDR,Map matching is a useful way to improve the performance of PDR.The wall crossing detection of Map Matching need involve a particle filter to process the measurements.This improve the accuracy of indoor localization.This thesis proposes partitioning rooms to improve the complexity.Additionly,integrating PDR into the map supports the navigation for users.