Study Of Inertial Pedestrian Positioning System Based On Building Structures

Inertial pedestrian navigation system is a must supplement for positioning in situations during GPS or other navigation signal outages. These situations include Search And Rescue(SAR) and Anti-terrorism missions in cities, mountain districts and forest zones. Inertial pedestrian navigation systems are able to work without any pre-installed infrastructures while it merely relies on self-measured 6-DOF acceleration and angular rate to perform positioning. During the trajectory-generation phase, classical Zero velocity UPda Te-aided Extended Kalman Filter(ZUPT-aided EKF) algorithm is exploited to suppress drifts brought by the Inertial Measurement Unit(IMU). However, the accuracy of the initially generated trajectory is still unacceptable in many applications, as a result, some calibration process should be introduced to increase location accuracy.Commonly used aiding knowledge to calibrate trajectory is building structures, because they can be employed to confine the movement of the pedestrians. This thesis mainly studies the features in building structures and how they can be exploited to perform trajectory calibration. The features of building structures are five folds, they are domain direction, anchors, maps, floor specifications and human walking behaviors respectively. These features are extracted from buildings and then expressed in unique ways to help calibrations. During the calibration process, a particle framework is specially proposed for convenience exploitation of domain directions, maps and human walking behaviors. In the end of each specification of method, sufficient experimental data are provided to validate their effectiveness in increasing location accuracy.The innovation points of this thesis lies in four aspects: 1) Human walking behavior is firstly combined with domain directions in buildings to adaptively adjust particle weights in the particle filter. In an outdoor complex-trajectory walking test of 908 meters' total walking distance, the location error is less than 0.5% of walking distance. Compared with raw trajectory, this accuracy has been significantly improved. Also, this method has better robustness than the HDE method, which may result in wrong heading estimation in a complex walking trajectory; 2) A revised method for map-aided pedestrian navigation is proposed, which can detect the number of cross-wall particles and thus having the advantage of wrong map tolerance. On the basis of traditional binary weighting of particles, a direction-oriented weighting strategy is added. The location error can maintain below 1 meter in a random walk; 3) An approach for height error suppression based on apriori knowledge is proposed, which focuses on long-term height drift suppression along with proper floor determination. This method is proposed under the assumption that a person's height only changes on stairs in the building. Based on this assumption, the whole course of walking in a building is partitioned into floor phase and stair phase, according to a built HMM model. For the floor phase, the height remains consistent on the same floor, and minor drifts are suppressed. For the stair phase, after sufficient stair-related information is acquired, a Maximum Likelihood Estimation(MLE) method is used to estimate the height of each stair step, and drift-free height difference between two floors can be estimated using the estimated stair step height. The experimental results of the proposed approach demonstrate the effectiveness of the approach for height error suppression and accurate floor determination. 4) A novel anchor based approach is proposed, which define anchor as the representative locations in buildings. Based on the invariance feature of these anchors, trajectories can be revised. Under the condition of frequent revisits, the approach has significant effects in error suppression.