A Novel Low-Cost UWB/IMU Positioning Method With the Robust Unscented Kalman Filter Based on Maximum Correntropy
Xin Li, Junhua Ye, Ze Zhang, Fei Liang, Wang Xia, Bo Wang
Abstract
Low-cost ultrawideband (UWB)/inertial measurement unit (IMU) integrated positioning faces performance degradation in complex non-Gaussian (NG) environments. Currently, the maximum correntropy (MC) filter has shown certain superiority in NG environments. In view of the problems of time-consuming, low adaptability, and numerical instability which are common in the current MC filtering, this article proposes a novel MC-based low-cost tightly-coupled (TC) UWB/IMU positioning method with the robust unscented Kalman filter (UKF), namely, MC-RUKF. Specifically, we propose an adaptive calculation method for MC Gaussian variable kernel bandwidth (VKB) based on UWB innovation (MC-VBK). Subsequently, the UWB measurement noise matrix can be adaptively calculated, referred to as the MC-VKB-R model, which can dynamically adjust the weight of UWB observations, aiming to adapt to more complex observation environments. Experimental results from three trajectory sets show that the MC-RUKF algorithm achieves optimal positioning accuracy in both Gaussian and NG environments. Moreover, the proposed MC-VKB method effectively reduces the impact of UWB abnormalities in UWB/IMU positioning and addresses the numerical instability issues associated with the traditional MC fixed kernel bandwidth (MC-FKB).