For multi-packet described linear discrete uncertain system, when the state can not be measured, an direct
method of constrained robust model predictive control algorithm is proposed. The observer and controller are designed
comprehensively, and observed state is used to structure performance index directly. Through solving the min-max
optimization problem with infinite horizon performance cost, the system optimal state feedback control law is obtained.
When the constrains of input/state is met, the parameter-dependent Lyapunov function are introduced to ensure the closed-
loop system stability. Simulation examples show the effectiveness of the proposed approach.