Assigments from 5LMB0 - Model Predictive Control course in TU EIndhoven for 2021/2022 academic year. The task is to implement a robust MPC, from linear MPC up to LPV-MPC to control the non-linear dynamics. The Electrically-driven Inverted Pendulum is based on Section 8.5.2 page 171 from Hanema J., Anticipative MPC for LPV, 2018. This implementation requires the additional MPT3 toolbox and YALMIP. For nicer implementations, one can also try to use the GUROBI inside the YALMIP as well. We managed to do this, however, it gives some crashes to my colleague's setup, thus it is recommended for the LPV-MPV to use the built-in YALMIP optimizers.
- Graded Assignment 1
This assignment is to linearise the plant so that one can apply a linear MPC to control the pendulum's trajectory.
- Graded Assignment 2
To obtain stability from MPC on the first assignment, one can use the Lyapunov's stability so that the stage cost can be maximised for stability in Part 1. Other way around, one can also use LMI (Linear Matrix Inequalities) method to actually compute all of the components based on the dARE equations in part 2
- Graded Assignment 3
This is the part where one can control the nonlinear dynamics using the LPV formulations. Here, one can run the first the initialDataGenerator to obtain the Uk-en-Rho.mat file. Afterwards, runing the Simulink Simulation or the [Closed-Loop Simulation](./Graded%20Project%203/assignment3_ClosedLoop.m for Simulink) and closed-loop simulations respectively. All of the results were adjusted so every states converges to zero. Otherwise, as they go to a constant, they are already at the local minimas.