Model Predictive Control
Towards Frontier Applications in Learning and Control 

This PhD course explores the fundamental ideas and principles of Model Predictive Control, nowadays regarded as one of the most popular and powerful control strategies.  We aim to familiarize students with the theoretical and mathematical foundations on which MPC is built, its strengths and weaknesses, as well as its computational aspects. This will serve as a starting point for exploring the many variants of MPC, including data-driven and reinforcement learning-based ones.

By taking this course the student will learn to:

• explain the fundamental principles of model predictive control;

• autonomously explore the tangle of MPC approaches, choosing the one best suited to the specific situation;

• implement an MPC controller to solve a challenging control problem.

Examination

Students will undertake a course project in groups (2-3 people/group), where they get to apply MPC to a real-world problem they deem interesting. The output of this project will be

• A short technical report of their project and reproducible code.

• A 15-minutes presentation of their project to their colleagues.

Basic exercises or quizzes will be given throughout the course. Attending Lecture 9 is, in general, mandatory.

Recommended prerequesites

• Familiarity with linear algebra, differential equations, and optimization

• Programming experience with MATLAB, Python, or Julia

• Basic knowledge of control theory (e.g., Automatic Control I-II)

Teachers

• Per Mattsson, Associate Professor at the Uppsala University

• Fabio Bonassi, Postdoc at the Uppsala University

• Daniel Arnström, Postdoc at the Uppsala University

Questions?

Contact Fabio Bonassi.