Model Predictive Control

I am a teaching assistant for the EL2700 Model Predictive Control (MPC) course at KTH, examined by Prof. Mikael Johansson. My main responsibility in the course is on creating and implementing MPC design assignments (a total of 4 assignments), and design the final project.

My main contributions to the course are:

1. Move the assignments from MATLAB to Python, and use CasADi to generate the the optimization problems that are then used in real-time for control;
2. Create the new assignments based on the Quadrotor Dynamics (2020) and the NASA Astrobee free-flyer (2021, 2022). During the course, the students solve problems encompassing pole-placement, finite-time optimal control, Linear Quadratic Regulator and Model Predictive Control.

For access to the assignment material, feel free to contact me or Prof. Mikael Johansson.

The course terminates with the final project demonstration. On the last two years, the demonstration session started with a presentation from someone who works/worked with the NASA Astrobee free-flyer (Brian Coltin in 2021, Keenan Albee in 2022, and Marina Moreira in 2023), followed by a simulated “ISS Test Session”, where we replicate what happens on a real test-session with astronauts.

First, the students prepare their project and must submit it by the pre-specified deadline. Once the code is submitted, it is also code-frozen - that is, no additional changes are permitted. Then, the code of each group is tested on a rendezvous task, and it’s performance is evaluated against the peers. The best performing group received a symbolic prize!

Below, you can see the final project demonstration from 2021 and 2022.