This repository was developed by Edoardo Barba and Matteo Brugnera and it contains the code and documents for the final project of Optimization Robot Control course of the Master in Artificial Intelligent Systems at the University of Trento, a.y. 2022-2023.
We delve into the problem of optimal control and stabilization of both the pendulum and double pendulum systems. The pendulum system, consisting of a mass attached to a fixed pivot point, exhibits complex dynamics that require sophisticated control strategies to achieve desired objectives. In this specific scenario, our objective is to bring the pendulum to an upright position. We achieved this result by utilizing the powerful Deep Q-Network (DQN) algorithm. The DQN algorithm has gained substantial popularity due to its ability to learn control policies directly from raw sensory inputs, making it an ideal choice for addressing complex control problems such as this one.
More details can be found in the project report.