An approximation of the Bayesian state observer with Markov chain Monte Carlo propagation stage

The state estimation problem for nonlinear systems with stochastic uncertainties can be formulated in the Bayesian framework, where objective is to replace completely by its probability density function. Without restriction selected system classes and disturbance properties, estimator particularly interesting highly non-Gaussian noise. main limitations of filters are significant computational costs implementation problems higher dimensional systems. present paper introduces a piecewise linear approximation observer Markov chain Monte Carlo propagation stage kernel estimation. These methods suitable prediction multivariate functions. proposed algorithms increase performance at reasonable cost. demonstrated benchmark comparing an extended Kalman filter particle filter.