Study On Smartphone-based Self-contained Pedestrian Localization Algorithm

Localization Based Service(LBS)plays a major role in people's life.Smartphone users can easily obtain the location information through GPS,but the system relies on the satellite assistance.In indoor environment,satellite positioning system will be influenced by signal blocking and positioning failure.Autonomous pedestrian positioning technology can achieve localization function without external auxiliary equipment,which has great research value.This thesis investigates a self-contained pedestrian localization algorithm based on smartphone,which is implemented in the framework of Pedestrian Dead Reckoning(PDR)algorithm.The main research addresses the following three issues:(1)The PDR algorithm is studied and its sub-algorithm is improved.The proposed method analyzes the gait characteristics of pedestrians,extracts the key events as stride detection conditions,and improves the performance of stride detection algorithm.The sensor fusion method based on gradient descent is improved.Heading drift is suppressed and the filtering divergence caused by magnetic field interference is avoided.A stride length estimation model is determined by using the maximum likelihood estimation method to identify the parameters.(2)Indoor map matching based on particle filter is realized on the basis of PDR,which effectively addresses the issue of pedestrian's localization and track passing the wall.This method improves the pedestrian positioning accuracy and optimizes the positioning trajectory,making the filtered trajectory more consistent with the ground truth.(3)A number of experiments are carried out to verify the performance of such algorithms as stride detection algorithm,heading estimation algorithm and stride length estimation algorithm.The quantitative index of localization results is given and analyzed.To verify the effectiveness and accuracy of the algorithm,experiments are performed in indoor and outdoor environments.Relative localization errors of the two trajectories in an outdoor stadium are 0.30% and 0.41%,separately.The mean relative position errors of three kinds of indoor walking tests are 1.62%,1.00% and 0.92%,respectively.After the map matching based on particle filter,the relative position error is reduced from 1.60% to 0.45% and 0.51%,respectively.The results demonstrate that the proposed algorithm has higher accuracy.