Square-root Unscented Kalman Filter of Small UAV Integrated Navigation System

曲藴杰,孙尧,莫宏伟
DOI: https://doi.org/10.3969/j.issn.1673-9728.2009.01.026
2009-01-01
Abstract:The process models and measurement models of the integrated navigation system,which is applied to small UAV,are strongly nonlinear.When the effects of the higher order terms of the Taylor series expansion become significant,EKF can seriously affect the accuracy or even lead to divergence of any inference system.A new filter algorithm based on square root unscented Kaman filter(SRUKF) is presented and used to the integrated navigation system of small UAV.SRUKF takes covariance square root matrix instead of covariance one in filter iteration,which effectively avoided filtering divergence,and meanwhile improved the convergence velocity and stability of the algorithm.The results of simulation prove that SRUKF is a more ideal nonlinear filtering method for the integrated navigation of small UAV and can obtain higher accuracy solutions.
What problem does this paper attempt to address?