Agile Bayesian filtering
Huazhen Fang, Xin Chao Zhao, Raymond A. DE CALLAFON · 2013
A novel nonlinear filtering approach, the agile Bayesian filter, is presented in this paper. Its design is directly based on the Bayesian filtering paradigm, a framework particularly useful for development of nonlinear filters. Compared to some existing filters, the agile Bayesian filter is less reliant on the Gaussian distribution approximations, the use of which is common in nonlinear filtering studies but indeed difficult to be justified. The agile Bayesian filtering formulae involve several Gaussian weighted integrals that need to be evaluated for implementation. They are numerically solved by the Monte Carlo integration method and the obtained filter is named the Monte Carlo agile Bayesian filter. The proposed filter is investigated through a simulation based study. Future improvements to this filter can be performed by using more accurate numeric integration rules.