Stochastic MPC Using the Unscented Transform

Model predictive control for linear stochastic systems with constraints has previously been solved using so called scenario methods which gives an approximate solution which is very computationally demanding. Here the model predictive control problem for linear stochastic system with known fixed system matrices are considered. Only outputs with measurement noise is assumed measurable. The main contribution is the development of a deterministic convex standard MPC problem which approximate the solution to the stochastic MPC problem.