Abstract This article deals with the estimation of a state of a system described by a nonlinear and non-Gaussian state-space model. The focus is on grid-based filters, which provides robust yet efficient numerical solution to the Bayesian recursive relations in a grid of deterministically selected points. In particular, the stress is laid on the survey of techniques for (i) grid design, (ii) computational and memory requirements reduction, and (iii) estimation performance improvement. The performance of the grid-based filters is illustrated in a highly nonlinear terrain-aided navigation based on real dataset. Data and the MATLAB® source codes are published along with the article.
Matoušek et al. (Tue,) studied this question.