Second-Order Simplex Sigma Points for Nonlinear Estimation
Jean-François Lévesque · AIAA Guidance, Navigation, and Control Conference and Exhibit · 2006
This paper proposes a novel technique for the selection of the sigma points used in Unscented Kalman Filters. It investigates a second-order algorithm that matches the moments of the probability density function up to the sixth order with a reduced set of sigma points, hence smaller processing requirements for increased estimation accuracy. Thus, an explicit second-order sigma-point solution is developed which, for the n-dimensional space, requires only 2 n+3 sigma points. Finally, the performance of the algorithm is compared to other techniques in the literature through simulations.