Weighted measurement fusion filter for multi-sensor uncertain systems with time-correlated measurement noises*

This paper is concerned with the globally optimal fusion estimation problem for multi-sensor stochastic uncertain systems with time-correlated measurement noises. First, the considered system is transformed into a normal system with uncorrelated white noises. State augmentation and measurement difference are the key technologies. Then based on full-rank decomposition and weighted least-squares method, the weighted measurement fusion filter is proposed in the sense of the linear minimum variance. The proposed weighted measurement fusion filter has lower computational cost and the same accuracy as the centralized one for the same initial values. It is globally optimal. A tracking system example shows the effectiveness of the proposed algorithm.