Abstract: High-performance control of unmanned surface vehicle (USV) trajectory tracking requires the systems to obtain motion state and uncertainty with high precision. However, the lack of direct highly precise measuring methods and the strong nonlinearity and coupling make it difficult to obtain the information needed. As to solve this problem, a kind of novel estimation algorithm is proposed, which combines singular value decomposition unscented Kalman filter (SVDUKF) with acceleration measurement together. SVDUKF is an improved algorithm of unscented Kalman filter (UKF) with wider application conditions. In addition, its main advantage lies in binding the ability of UKF to deal the strong nonlinearity with the feather of acceleration with much disturbance information to simplify the system estimation model so that it becomes an online estimation algorithm with higher precision and lower calculation complexity. The nonlinear model of USV is derived, and the model simplification idea of acceleration and the basic steps of SVDUKF are introduced. The advantages in estimation accuracy and computation efficiency of the proposed algorithm are verified.

Key words: Acceleration, Computational complexity, State estimation, Unscented Kalman filter