A dual controller for linear systems with random jump parameters

Abstract
This paper addresses the problem of control of partially observed linear discrete time stochastic systems with unknown parameters taking values in a finite set. The solution approach is based on a decomposition of the cost function into a control and trajectory component and a dual or learning component. Similarly the control policy is decomposed into a trajectory and a dual policy. A novel identification feedback scheme which introduces duality into the problem is presented.

This publication has 14 references indexed in Scilit: