Robust Kalman Filter Algorithm Based on Generalized Correntropy for Ultra-Wideband Ranging in Industrial Environment

Fuqiang Ma,Jie He,Xiaotong Zhang

DOI: https://doi.org/10.1109/access.2019.2901940

IF: 3.9

2019-01-01

IEEE Access

Abstract:The performance of time-of-arrival (TOA)-based ranging using ultra-wideband is greatly declined in the industrial environment since the metallic obstacles cause severe non-line-of-sight (NLOS) and result in huge ranging measurement errors. A general challenge of TOA-based ranging and localization in the industrial environment is that the Kalman filter (KF)-based ranging optimization algorithm cannot effectively improve the ranging accuracy because the ranging errors follow a non-Gaussian distribution. In this paper, a generalized maximum correntropy Kalman filter (GMCKF) algorithm which can effectively suppress NLOS errors is proposed. GMCKF uses the generalized maximum correntropy criterion (GMCC) instead of the minimum mean square error as the criterion of KF, and obtain a robust gain function. GMCC can effectively measure the similarity between the state value and the measurement value, which directly reflects the abnormality of measurement errors. Therefore, GMCKF achieves smoothing filtering in both NLOS and line-of-sight conditions. We compare GMCKF with other KF-based algorithms and prove its steady-state performance in field testing. The results show that the ranging optimized by GMCKF is with significantly higher accuracy. Finally, the optimized ranging is used as the input of three general localization algorithms. The localization accuracy of all localization algorithms is also improved.