Model Predictive Control refers to a class of control methods where the model of the process is used to obtain the control signal up to \(N\) timesteps in the future such that there is a duration of \(dt\) seconds between each timestep. The control signal is obtained by minimizing an objective function while satisfying a set of constraints and only the first control is used after which we optimize again for the next \(N\) timesteps. Usually the main differences between the various MPC algorithms are in the model used to represent the process, the cost function and the set of constraints.

The MPC consists of two important blocks:

  1. The model: The model used to represent the process must be able to capture the dynamics of the process while being simple enough to be implemented and understand.
  2. The optimizer: The optimizer used to minimise the cost function also plays an important role in getting the optimal controls and the quality of the controls also depends on the optimizer used and the formulation of the cost function.


  1. [MPC in Autonomous Vehicles]


  1. Camacho, Eduardo F., and Carlos Bordons Alba. Model predictive control. Springer science & business media, 2013.