Adaptive robust INS/UWB-integrated human tracking using UFIR filter bank

For human navigation, the ultra wide band (UWB)-based positioning technologies require extra infrastructure, while the inertial navigation system (INS)-based ones produce drift errors on a long time-scale. To overcome these shortcomings, we propose a novel INS/UWB-integrated system, in which the difference between the INS-measured position and UWB-measured position is corrected using a robust unbiased finite impulse response (UFIR) filter in a fusion center. To estimate optimally the averaging horizon for the UFIR filter in real time, the UFIR filter bank (UFFB) algorithm is designed employing the Mahalanobis distance between the estimated measurement and the actual measurement. It is shown experimentally that the UFFB-based INS/UWB-integrated human tracking system is able to provide real-time estimation with an accuracy consistent to that of the UFIR filter, which relies on a constant optimal horizon.