F Discrepancy Rep-points Based Gaussian Process Moment Matching Filter
Yao Zhi · Yuhang jice jishu · 2011
To solve the filtering problem with unknown system models,a F-discrepancy rep-points based Gaussian process moment matching filter is advanced in this paper.This filter learns the transition and observation models by Gaussian processes and then computes numerically in the structure of Gaussian moment matching filtering with F-discrepancy rep-points.Simulations in both single variants nonlinear system and the initial alignment problem have show that the filtering precision of this F-discrepancy rep-points based Gaussian process filter is not lower than GP-ADF and much better than GP-UKF.