Variational Bayesian Kalman filter using natural gradient |
| |
Institution: | 1. School of Automation, Northwestern Polytechnical University, Xi’an 710072, China;2. The Key Laboratory of Information Fusion Technology, Ministry of Education, Xi’an 710072, China;3. Department of Electrical and Electronic Engineering, The University of Melbourne, Melbourne, VIC 3010, Australia;4. School of Engineering, RMIT University, Melbourne, VIC 3000, Australia;5. School of Computer and Information Engineering, Henan University, Kaifeng 475001, China |
| |
Abstract: | We propose a technique based on the natural gradient method for variational lower bound maximization for a variational Bayesian Kalman filter. The natural gradient approach is applied to the Kullback-Leibler divergence between the parameterized variational distribution and the posterior density of interest. Using a Gaussian assumption for the parametrized variational distribution, we obtain a closed-form iterative procedure for the Kullback-Leibler divergence minimization, producing estimates of the variational hyper-parameters of state estimation and the associated error covariance. Simulation results in both a Doppler radar tracking scenario and a bearing-only tracking scenario are presented, showing that the proposed natural gradient method outperforms existing methods which are based on other linearization techniques in terms of tracking accuracy. |
| |
Keywords: | Kullback-Leibler divergence Natural gradient Nonlinear Kalman filter Target tracking Variational Bayesian optimization |
本文献已被 ScienceDirect 等数据库收录! |
|