Neural Networks for Optimal Control

Exam format

- Upon taking an exam, the course gives 3 ECTS for EECI students and 2 ECTS for EPFL students 

- The exam will be based on a project to be developed after the course, to be handed in before Friday, May 16th. The project can be done individually or in groups of up to 4 people.

Project contents

The goal is to apply a control technique based on neural networks seen in the course to a system of your choice and conduct simulations.

1. Choose a dynamical system to be controlled and, if necessary, design a pre-stabilizing controller. You can pick any system you like. You can also take a system from COMPlib - see here for a description of the available models.
2. Define a meaningful cost to be optimized to reach desired control objectives.
   
3. Solve the corresponding optimal control problem using one of the neural network models seen in the course.
4. Write a short report (max 10 pages) describing the system and control design procedure you have followed and providing simulations showing the usefulness of the neural network controller.
5. Send the report by email to giancarlo.ferraritrecate@epfl.ch