Robust Output Feedback MPC for Networked Control Systems with Two-Channel Random Packet Dropouts.

The article addresses the design of a robust output feedback Model Predictive Control (MPC) scheme for linear constrained Networked Control Systems (NCSs) experiencing disturbances, observation noise, and random packet dropouts in both Sensor-Controller (S-C) and Controller-Actuator (C-A) communication channels. By extending the Luenberger observer to a switched system framework, the authors develop Generalized Robust Positive Invariant (GRPI) sets that explicitly bound estimation and prediction errors caused by packet dropouts and disturbances. These GRPI sets enable the formulation of tightened constraints within the MPC optimization problem, ensuring robust satisfaction of both state and input constraints. The proposed control scheme is proven to be recursively feasible and guarantees closed-loop stability, with simulation results on a quadrotor model validating its effectiveness under significant packet dropout scenarios.