Alternative framework for the iterated unscented Kalman filter. Issue 3 (1st May 2017)

• Record Type: Journal Article
• Authors: Chang, Guobin; Xu, Tianhe; Wang, Qianxin
• Is Part Of: IET signal processing. Volume 11:Issue 3(2017)
• Page Start: 258
• Page End: 264
• Publication Date: 2017-05-01
• Subjects: nonlinear filters -- Kalman filters -- regression analysis -- iterative methods; alternative framework -- iterated unscented Kalman filter -- nonlinear Kalman filters -- unscented transform -- UT‐based UKF -- statistical linear regression perspective -- iterated extended Kalman filter -- IEKF -- observation equation -- SLR perspective -- regression error -- linearised equation -- a priori estimate; Signal processing -- Periodicals; 621.3822
• DOI: 10.1049/iet-spr.2016.0214
• Languages: English
• ISSNs: 1751-9675

Abstract

Abstract : The iterated version of a family of non‐linear Kalman filters, named the unscented transform (UT) based unscented Kalman filters (UKF), are revisited. Two existing frameworks of the iterated UKF are analysed and some shortcomings of them are pointed out. A new framework is proposed based on the statistical linear regression (SLR) perspective of the UT and the framework of the iterated extended Kalman filter (IEKF). The virtue of the proposed framework is twofold: first, the observation equation is linearised strictly following the SLR perspective implying that the regression error is also considered; second, it strictly follows the framework of the IEKF implying that in each iteration, the linearised equation is used to correct the a priori estimate rather than the latest estimate. A simple but illustrative benchmark example is simulated to check the feasibility of the proposed framework, and the results demonstrate the efficacy of the proposed framework.