Near-optimal control of discrete-time nonlinear stochastic systems

The problem of finding a nonrandomised control law that minimises an instantaneous index for a nonlinear discrete-time stochastic system is considered. Both the state and observation equations are taken to be nonlinear, and the optimisation index is chosen to be the variance of the control error. Some of the recent results in nonlinear estimation theory are utilised for deriving approximations to the optimal-control law. The results are illustrated with the help of numerical examples.