Distance Constraint Ensemble Kalman Filter for Pedestrian Localization

To enhance the positioning accuracy of the inertial measurement unit (IMU)-based pedestrian localization, this study proposes an adaptive ensemble extended Kalman filter (EnEKF) that incorporates a distance constraint (DC). This study first introduces a dual foot-mounted IMU-based pedestrian localization system that employs two IMUs to measure the target human's position. Second, an augmented data fusion model is developed by incorporating attitude quaternions from the inertial navigation system (INS) into the conventional INS error-state vector. Based on this new data fusion model, a DC-based EnEKF is designed. In this method, the EnEKF employs ensemble factors to address nonlinear and non-Gaussian characteristics inherent in the data fusion process. Then, the colored measurement noise (CMN) is considered, and the method is modified to form an EnEKF under CMN (cEnEKF). Moreover, the DC is employed to further restrict the INS-derived position estimates of the left and right feet obtained from the EnEKF algorithm. Finally, validation in two real-world scenarios confirms the effectiveness and superior performance of the proposed approach.