Predictive control with partial state information

A method is described in order to address control problems in the presence of state and input constraints for noisy linear plants under partial state information. The approach consists in reducing the problem to a full-state feedback strategy for an auxiliary system subject to suitably modified constraints and disturbances. This is particularly useful in the context of model predictive control since traditional MPC schemes usually require exact knowledge of the state. Moreover, the overall procedure leads to MPC schemes based on closed-loop predictions and hence less affected by conservativeness. As an example of application we discuss the implementation of a predictive controller for a two-carts/spring system under input constraints and bounded persistent disturbances.