MATLAB Answers


Please recommend resources about applications of reinforcement learning on control problems

Asked by jhusy
on 9 Aug 2019
Latest activity Edited by jhusy
on 22 Aug 2019
Hi everyone, I'm a student majoring in control engineering. Recently my advisor asks me to use reinforcement learning to tune a motor's controller parameters (like PID or feedforward controller's parameters), but I have no idea how to do.
I've already watched videos like Matlab's Tech Talks and have a basic understanding of how machine learning and one of its branches, reinforcement learning, work. However, I've never actually implemented these algorithms by myself before, nor do I have rigorous theoretical and mathematical foundations on ML & RL.
I know there's a watertank example in Matlab, but what it does is replacing the whole PID controller with a RL agent, instead of using RL to tune PID controller's parameters, making it difficult to be modified for my purpose.
I may take complete courses on ML & RL after the next semester begins, but for now I need to get to understand HOW to use RL (no matter using RL toolbox or code by myself) instead of RL itself.
My questions are below:
  1. Are there resources (books or courses or anything) on using RL on control problems with Matlab & Simulink? I’ve found plenty resources on RL, but they are all in Python and focusing on CS-related subjects. I need to use Simulink to obtain simulation results of my system.
  2. Are there online courses that are short and comprehensible for a non-CS background student? So I can have a quick overview on what I’m doing.
Thanks for any suggestions in advance.


Sign in to comment.

1 Answer

Answer by Harsha Priya Daggubati on 12 Aug 2019
 Accepted Answer

You can refer to the following links:
  1. Video Tutorials on Reinforcement Learning, to help you gain an intuition about the topic.
  3. Documentation Link along with Tutorials.
  5. About Reinforcement Learning Toolbox in MATLAB.

  1 Comment

Though it's not what I expected, still thanks anyway!!

Sign in to comment.