Inverse Minimax Optimality of Model Predictive Control Policies

We present results linking model predictive control (MPC) and minimax optimal control theory. A distinction from previous work is that we show that typical MPC policies, which use the current system state, are minimax optimal closed-loop policies with respect to a certain class of cost functions. The control algorithms under study are similar, and sometimes identical, to conventional MPC. They require the solution of a single quadratic programming problem in each step