Model-free RL Algorithms Implementation

About

In the course CS4400 Deep Reinforcement Learning at Delft University of Technology, I implemented the main modern model-free deep reinforcement learning (DRL) algorithms in the course assignments. I found the theory and math behind very beautiful and want to share (discuss) its core ideas and future applications in robotics, especially combined with traditional methods. Therefore, I am writing some blogs in my free time. Besides, I am also working on an open source project where I implemented some small-scale (and thus for fast prototyping and baseline usage) RL algorithm implementations using the programming language Julia.

Proudly, I wrote an exam with a grade of 10/10 for the course (top 1 in the class) and will serve as a teaching assistant for the 2022/23 academic year. Welcome to take the course and learn the beauty of DRL theory!

Supplementary Video

One of the videos that shows the tricks (the introduction text below the video) for stabilizing the learning of value functions. Due to the restriction, I am not allowed to publish the code. But I will write more blogs in the near future!