Fourier Subspace Preserving Particle Filter (Poster)

Abstract

We introduce a new particle filter, based on the mathematical notion of weak approximation. The filter is similar to the Daum-Huang particle filter in that it propagates filter particles based on a “homotopy” that is used to adjust the position of particles. However, the homotopy that we propose is much simpler, and does not require extraneous mathematical constructions. We verify the filter's performance in one and two dimensions. The filter holds promise of providing a low-complexity, high performance filter.