lambert-x/ReinforcementLearning-DQN-MPC

Course Project of Reinforcement Learning

Reinforcement Learning Course Project

Technische Universität Darmstadt winter semester 2018/2019

Supervisor: Jan Peters, Riad Akrour

This repository contains the PyTorch implementation of Deep Q-Network and Model Predictive Control (MPC), and the evaluation of them on the quanser robot platform.

Authors

• Zuxin Liu ,Yunhao Li, Junfei Xiao

Algorithms

• DQN
• MPC

Platforms

• Qube
• Double Pendlum
• Cartpole Swing-up
• Cartpole Stab

Installation

For the installation of the Quanser robot simulation environment, please see this page

For the implementation of the algorithms, the following packages are required:

• python = 3.6.2
• pytorch = 1.0.1
• numpy = 1.12.1
• matplotlib = 2.1.1
• gym

You can simply create the same environment as ours by using Anaconda. All the required packages are included in the environment.yaml file. You can create the environment by the following command

conda env create -f environment.yaml

Then, activate your environment by

source activate pytorch

How to run

1. Choose the algorithm you want to use and change to the corresponding folder (DQN or MPC)
2. Choose the environment you want to evaluate and change to the folder (CartPoleStab, Double, Qube or Swing)
3. Change the configuration file config.yml to the parameters you want, and follow the instructions in the folder