Multiple sensor estimation using the sparse Gauss-Hermite quadrature information filter

Abstract

In this paper, a sparse Gauss-Hermite quadrature information filter (SGHQIF) is proposed for multiple sensor estimation. The new proposed information filter is more flexible to use and can achieve higher level estimation accuracy than the extended information filter and the unscented information filter. In addition, the new filter maintains the close performance to the conventional Gauss-Hermite information filter with significantly fewer quadrature points and is thus computationally more efficient. The performance of these information filters is compared via a target tracking problem and the SGHQIF is shown to be the best one balancing the estimation accuracy with computational efficiency.