Skip to Main Content
Through much research on unscented Kalman filter based on scaled and square-root algorithm in autonomous navigation, it is known that these algorithms take so much time on calculation. So an improved unscented Kalman filter algorithm is proposed in the paper for autonomous navigation to solve the non-real-time difficulty. Simple scheme is adopted to predigest the choose procedure of sigma-points and weights, which reduces a mass of complex operations. From both theory and practical simulation, it is shown that much mass of calculation is reduced when the dimension of state matrix is large, and not leads to bad filtering performance.