NONLINEAR FILTERING TOOLBOX FOR CONTINUOUS STOCHASTIC SYSTEMS WITH DISCRETE MEASUREMENTS

The paper deals with a problem of state estimation for nonlinear continuous stochastic systems with discrete-time measurements. A general recursive solution of the estimation problem given by the Bayesian rule and by the Fokker-Planck equation is described. Local estimation methods employing analytical approach to solution and global estimation methods employing analytical, numerical and simulation approaches are discussed. A software package for state estimation of continuous stochastic systems with discrete-time measurements is developed and described. It serves for system design, system simulation, estimator setup and state estimation. The package is designed to embody easily user defined estimators and thus it is suitable for estimator testing and quality comparison of different estimators. Usage of the nonlinear filtering software package is illustrated in a numerical example.