An Adaptive Variational Outlier-Robust Filter for Multisensor Distributed Fusion

The fusion accuracy of distributed fusion algorithms may degrade in multisensor measurements with unknown noise and outliers. To tackle the problem, an improved adaptive feedback covariance intersection (IAFCI) fusion algorithm based on the modified slide window variational outlier-robust adaptive Kalman filter (MSWVRAKF) is proposed. To estimate unknown noise covariance and eliminate outliers, an MSWVRAKF is proposed. The algorithm devises a simplified slide-window variational adaptive filter according to the Student’s t distribution, which treats the Student’s t distribution as the approximation of the posterior distribution to eliminate the effect of outliers. The adaptive factor is introduced into this algorithm to realize the tradeoff between measurement and prediction. Moreover, an IAFCI fusion algorithm is developed with respect to multisensor information fusion with uncertain noise. The simulations verify that the improved fusion algorithm outperforms other existing filtering and fusion algorithms.