1. Institute of Advanced Digital Technology and Instrumentation, Zhejiang University, Hangzhou 310027, China; 2. Zhejiang Provincial Key Laboratory for Network Multimedia Technologies, Hangzhou 310027, China
A robust Kalman filtering algorithm with estimation of measurement biases was proposed in order to handle the problem that unknown measurement biases and random measurement noises exist in the measurement system. The nonzero-mean Gaussian distribution model was utilized to model the measurement biases and the random measurement noises of the measurement system. The Normal-Inverse-Wishart distribution was utilized to estimate the mean and covariance parameters of the Gaussian distribution. The time-variant parameters of the mixed model between the Gaussian distribution and the Normal-Inverse-Wishart distribution were inferred by the variational Bayesian approximation. The measurement biases and the time-variant covariances of the random measurement noises were estimated as the system states were recursively inferred by the cubature Kalman filter. The proposed algorithms robustness to the measurement outliers was enhanced with the estimation of the measurement biases. The simulation results demonstrate that the proposed algorithm can also precisely estimate the measurement biases and enhance its robustness with the guarantee of the high state estimation precision．
［1］ WELCH G, BISHOP G. An introduction to Kalman filtering ［R］. Chapel Hill, NC: University of North Carolina, 2006．
［2］ RABINER L, JUANG B. An introduction to hidden Markov models ［J］. IEEE ASSP Magazine, 1986, 3(1): 416．
［3］ ZHU H, LEUNG H, HE Z. A variational Bayesian approach to robust sensor fusion based on Student-t distribution ［J］. Information Sciences, 2013, 221(0): 201-214．
［4］ JO-ANNE T, THEODOROU E, SCHAAL S. A Kalman filter for robust outlier detection ［C］∥ IEEE/RSJ International Conference on Intelligent Robots and Systems. San Diego: IEEE, 2007: 1514-1519．
［5］ SARKKA S, NUMMENMAA A. Recursive noise adaptive Kalman filtering by variational Bayesian approximations ［J］. IEEE Transactions on Automatic Control, 2009, 54(3): 596-600．
［6］ AGAMENNONI G, NIETO J I, NEBOT E M. An outlier-robust Kalman filter ［C］∥ IEEE International Conference on Robotics and Automation. Shanghai: IEEE, 2011: 1551-1558．
［7］ AGAMENNONI G, NIETO J I, NEBOT E M. Approximate inference in state-space models with heavy-tailed noise ［J］. IEEE Transactions on Signal Processing, 2012, 60(10): 5024-5037．
［8］ SARKKA S, HARTIKAINEN J. Non-linear noise adaptive Kalman filtering via variational Bayes ［C］∥ IEEE International Workshop on Machine Learning for Signal Processing. Southampton: IEEE, 2013: 16．
［9］ BEAL M J. Variational algorithms for approximate Bayesian inference ［D］. London: University of London, 2003．
［10］ OKELLO N, RISTIC B. Maximum likelihood registration for multiple dissimilar sensors ［J］. IEEE Transactions on Aerospace and Electronic Systems, 2003, 39(3): 1074-1083．
［11］ TAGHAVI E, THARMARASA R, KIRUBARAJAN T, et al. Bias estimation for practical distributed multiradar-multitarget tracking systems ［C］∥ 16th International Conference on Information Fusion. Istanbul: ISIF, 2013: 1304-1311．
［12］ GELMAN A, CARLIN J B, STERN H S, et al. Bayesian data analysis ［M］. Florida: CRC, 2013．
［13］ OZKAN E, SMIDL V, SAHA S, et al. Marginalized adaptive particle filtering for nonlinear models with unknown time-varying noise parameters ［J］. Automatica, 2013, 49(6): 1566-1575．
［14］ HERSHEY J R, OLSEN P A. Approximating the Kullback Leibler divergence between Gaussian mixture models ［C］∥ IEEE International Conference on Acoustics, Speech and Signal Processing. Hawaii: IEEE, 2007: 317-320．
［15］ ARASARATNAM I, HAYKIN S. Cubature Kalman filters ［J］. IEEE Transactions on Automatic Control, 2009, 54(6): 1254-1269．
［16］ COWLES M K. Applied Bayesian statistics ［M］. New York: Springer, 2013．
