SEPARATION PRINCIPLE IN OPTIMIZING CONTROL OF STATE-CONSTRAINED DYNAMIC-SYSTEMS UNDER BOUNDED UNCERTAINTY

Consideration is given to the problem of the design of a control law for a class of state-constrained dynamical discrete-time systems under bounded uncertainty. A set of recursive equations of a dynamic programming type is developed to characterize a closed-loop minimax control law. It is shown that a separation principle holds for the class of systems under consideration. The difference between stochastic and set-bounded models of uncertainty is pointed out in the context of the applicability of a separation principle.