/udacity-deep-reinforcement-learning-Collaboration-Competition

Train two agents to control rackets to bounce a ball over a net using Multi-Agent-reinforcement-learning. This is the third project of udacity-deep-reinforcement-learning nanodegree.

Primary LanguageJupyter Notebook

Project: Continuous Control

Introduction

This project is part of the Deep Reinforcement Learning Nanodegree Program, by Udacity.

In this environment, two agents control rackets to bounce a ball over a net. If an agent hits the ball over the net, it receives a reward of +0.1. If an agent lets a ball hit the ground or hits the ball out of bounds, it receives a reward of -0.01. Thus, the goal of each agent is to keep the ball in play.

The observation space consists of 8 variables corresponding to the position and velocity of the ball and racket. Each agent receives its own, local observation. Two continuous actions are available, corresponding to movement toward (or away from) the net, and jumping.

Getting Started

Dependencies

Follow the instructions in the DRLND GitHub repository to set up your Python environment. These instructions can be found in README.md at the root of the repository. By following these instructions, you will install PyTorch, the ML-Agents toolkit, and a few more Python packages required to complete the project. To set up your python environment to run the code in this repository, follow the instructions below.

  1. Create (and activate) a new environment with Python 3.6.

    • Linux or Mac:
    conda create --name drlnd python=3.6
source activate drlnd
    • Windows:
    conda create --name drlnd python=3.6 
activate drlnd

  2. Clone the repository 

    git clone https://github.com/udacity/deep-reinforcement-learning.git
cd deep-reinforcement-learning/python
pip install .

  3. Instructions
    Open the IPython notebook.

    jupyter notebook Continuous_Control.ipynb

    Before running code in a notebook, change the kernel to match the drlnd environment by using the drop-down Kernel menu.

    For this project, you will not need to install Unity - this is because we have already built the environment for you, and you can download it from one of the links below. You need only select the environment that matches your operating system. The Reacher.app is the environment app.

  4. Download the Unity environment with the Agents

    Download the environment from one of the links below and decompress the file into your project folder.
    You need only select the environment that matches your operating system:

    (For Windows users) Check out this link if you need help with determining whether your computer is running a 32-bit version or 64-bit version of the Windows operating system.

    (For AWS) If you'd like to train the agent on AWS (and have not enabled a virtual screen), then please use this link (version 1) or this link (version 2) to obtain the "headless" version of the environment. You will not be able to watch the agent without enabling a virtual screen, but you will be able to train the agent. (To watch the agent, you should follow the instructions to enable a virtual screen, and then download the environment for the Linux operating system above.)

Understanding the environment

The task is episodic, and in order to solve the environment, the agents must get an average score of +0.5 (over 100 consecutive episodes, after taking the maximum over both agents). Specifically,

  • After each episode, we add up the rewards that each agent received (without discounting), to get a score for each agent. This yields 2 (potentially different) scores. We then take the maximum of these 2 scores.
  • This yields a single score for each episode.

The environment is considered solved, when the average (over 100 episodes) of those scores is at least +0.5.

Solving the environment

The multi-agent DPG algorithm is used to solve the environment. The deep neural network is used to implement the actor and critic network. In each episodes, 1000 plays is conducted and the actor and critic network as well as the memory is shared between two agents. After tuning the parameters and the training procedures. The two agents can achive a score larger than 0.5 after about 100 episodes.