Sequential fusion filtering localization algorithm of unmanned surface vehicle based on multi-antenna measurement information

Abstract This paper investigates the fusion filtering algorithm for multi-antenna measurement information of an unmanned surface vehicle (USV) and proposes an optimal Kalman sequential fusion filtering (KSFF) algorithm with minimal linear covariance. The measurement information from each antenna on the USV is transmitted to the information fusion center and integrated in the order of reception. After fusing all the information, the state information of the USV based on global observations can be obtained and then updated for the next sampling period. Given the high computational complexity of KSFF, a measurement sequential fusion filtering (MSFF) algorithm with reduced computational complexity is introduced. The error covariance of the proposed algorithm has an upper bound, which has been demonstrated through mathematical induction, and the accuracy of state estimation is numerically equivalent to centralized fusion (CFF). Both simulation and physical experiment utilizing a USV are designed to verify that the KSFF and MSFF algorithm exhibit equivalent accuracy to the CFF algorithm. The experimental results show that the position and velocity accuracy of KSFF and MSFF algorithms are improved by 76.37% and 64.75%, respectively, compared with the results of single antenna. Additionally, a comparative analysis of the time consuming of the proposed algorithms is conducted, highlighting their computational advantages over the CFF algorithm.