Indoor Motion Trajectory Detection Using Buit-in Sensors In Smartphone

Nowadays, LBS (Location Based Service) is playing an increasingly important role, which has been applied into health care, safety protection, commercial advertising, intelligent travel, map navigation etc. Two key components in this domain are accurate location acquisition and motion trajectory estimation. As there exists signal reflection, multipath effect, in changeable indoor environment, traditional location methods based on external signal are far from satisfactory when put into indoor environment. INS (Inertial Navigation System) in smartphone, is a good alternative solution for indoor location service with advantages of the changeable location range, strong adaptation to environment and being independent of external signals. However, the inherent noise of built-in sensors, as well as the arbitrariness and inconstancy of both a pedestrian's motion state and his phone carrying state, limits its applicability, which will affect the accuracy and robustness in estimating motion trajectory. In this paper, first we design filters aiming at different scenarios. Second, we focus on estimating motion state and phone carrying state based on common features, which is inspired by "divide-and-conquer". Finally, based on the above results, we propose an accuracy method in estimating motion trajectory indoors with the assistance of indoor building structure, which is robust to motion state and phone carrying pattern.We mainly study:1. Noise eliminationFirst determine data composition of built-in sensors in smartphone through figuring out how they work, then detect features for different noise, and finally design targeted noise filters.2. Classification and Estimation of Motion StateFirst decouple motion state into three common feature states through analysis of people's motion characteristic, then figure out the strong related feature variable and an appropriate threshold for each feature state, finally recognize motion state via comparing and matching the qualitative incorporation.3. Classification and estimation of smartphone carrying stateFirst decouple phone carrying state into three common feature states through analysis of people's habits in using a smartphone, then figure out the strong related feature variable and an appropriate threshold for each feature state, finally recognize phone carrying state via comparing and matching the qualitative incorporation.4. Motion trail estimationBased on certain motion state and phone carrying state, a strong applicable to usage scenarios motion trajectory estimating method is proposed, which achieves high performance of accuracy and robustness.