Linear optimal estimation algorithms based on the Monte Carlo method and neural networks for nonlinear navigational problems

Problems of navigation data processing intended to estimate the time invariant vector by nonlinear measurements are considered. The Bayesian approach, which provides the basis for determining optimal (minimum variance) estimates, is used. Suboptimal algorithms based on linear optimal estimates are proposed. Two methods for calculation of these estimates are considered. One of them is based on the Monte Carlo method, the other uses neural networks. The algorithms proposed are compared with nonlinear optimal and linearized algorithms. An example and the results of application of linear optimal estimates to the problem of navigation with the use of reference beacons are given.