Robust model predictive control using tubes. (Q1426275)

A model predictive control (MPC) method for linear (time-varying) systems with bounded additive disturbances is suggested. Several asymptotically stable robust controllers which can steer any initial state of the actual system to a terminal constraint set at a given number of stages (along the trajectory lying in a tube) for all admissible disturbance sequences are specified. An algorithm for the construction of disturbance invariant tubes (using time-varying and piecewise affine control policy) is presented. Conditions to ensure that a tube controller satisfies all state, control, and terminal constraints are given. It is shown that the computational complexity of the proposed MPC is linear in horizon length. The suggested method is compared with several robust MPC approaches.