Solve Reacher Environment Using DDPG Agent

Introduction

In this project, we will create an agent to reach the green balooon. The agent is in form of a robotic arm which have a freedom to move in all direction. The term agent here is a robot built using Artificial Intelligence. In detail, we use Deep Deterministic Policy Gradient(DDPG) well-written in this paper.

We will use Reacher environment to train the agent. The following gif picture depicts the environment we will solve using the agent.

Keep in mind, our goal is move the agent's hand to touch the target baloon. A reward of +0.1 is provided for each step that the agent's hand is in the baloon location. Thus, the goal of your agent is to maintain its position at the target baloon for as many time steps as possible.

The Details on The Environment and The Agent

The observation space consists of 33 variables corresponding to position, rotation, velocity, and angular velocities of the arm. Each action is a vector with four numbers, corresponding to torque applicable to two joints. Every entry in the action vector should be a number between -1 and 1.

Distributed Training

For this project, we will provide you with two separate versions of the Unity environment:

• The first version contains a single agent.
• The second version contains 20 identical agents, each with its own copy of the environment.

The second version is useful for algorithms like PPO, A3C, and D4PG that use multiple (non-interacting, parallel) copies of the same agent to distribute the task of gathering experience.

Solving the Environment

Note that your project submission need only solve one of the two versions of the environment.

Option 1: Solve the First Version

The task is episodic, and in order to solve the environment, your agent must get an average score of +30 over 100 consecutive episodes.

Option 2: Solve the Second Version

The barrier for solving the second version of the environment is slightly different, to take into account the presence of many agents. In particular, your agents must get an average score of +30 (over 100 consecutive episodes, and over all 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 20 (potentially different) scores. We then take the average of these 20 scores.
• This yields an average score for each episode (where the average is over all 20 agents).

The environment is considered solved, when the average (over 100 episodes) of those average scores is at least +30.

Getting Started

Follow the instructions below to explore the environment on your own machine! You will also learn how to use the Python API to control your agent.

Step 1: Set Up The Depedencies

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. Follow the instructions in this repository to perform a minimal install of OpenAI gym.

   • Next, install the classic control environment group by following the instructions here.
   • Then, install the box2d environment group by following the instructions here.

3. Clone the repository (if you haven't already!), and navigate to the python/ folder. Then, install several dependencies.

git clone https://github.com/jhonsonlee/ddpg-in-reacher-env.git
cd ddpg-in-reacher-env/python
pip install .

4. Create an IPython kernel for the drlnd environment.

python -m ipykernel install --user --name drlnd --display-name "drlnd"

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

(For Windows users) The ML-Agents toolkit supports Windows 10. While it might be possible to run the ML-Agents toolkit using other versions of Windows, it has not been tested on other versions. Furthermore, the ML-Agents toolkit has not been tested on a Windows VM such as Bootcamp or Parallels.

Step 2: Download the Unity Environment

1. Download the environment from one of the links below. You need only select the environment that matches your operating system:

   • Version 1: One (1) Agent

     • Linux: click here
     • Mac OSX: click here
     • Windows (32-bit): click here
     • Windows (64-bit): click here

   • Version 2: Twenty (20) Agents

     • Linux: click here
     • Mac OSX: click here
     • Windows (32-bit): click here
     • Windows (64-bit): click here

   (For Windows users) Check out this link if you need help with determining if 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.)

2. Place the file in the ddpg-in-reacher-env repository.

Step 3: Getting Familiar with Jupyter Notebook

The code in this repository will be written mostly in .ipynb file. Thus, you need to get familiar with the jupyter notebook which is a great editor today. Please, take your time to know Jupyter Notebook in this blogpost.

Instructions

Navigating The Source Code

There are three important files which contains the source code to run and train our agent:

• Continious Control.ipynb contains the instructions to explore the environent, train the agent to be smart and run the smart agent.
• /src/agent.py contains the source code which describes how the agent works
• /src/model.py contains the source code of deep learning model for the DDPG algoritm.

Explore the Environment

In order to understand how to create a smart agent, we must recognize the environment first. Please, open Continious Control.ipynb using Jupyter Notebook and follow these steps:

• Step 1 : Start the Environment
• Step 2 : Examine the State and Action Spaces
• Step 3 : Take Random Actions in the Environment

Train the Agent

As you can see in the Explore the Environment section, the agent is still taking random actions and generate zero points during the gameplay. In order to create a smart agent, we need to train the agent by following these two steps:

• Step 4 : Contruct the Train Method
• Step 5 : Train the Agent

Watch the Smart Agent Play

After we train to agent how to better play the game, we can watch the agent play with better decision and better score. Howevever, we can not see the agent play in live gameplay in this notebook. But, we can see the agent collects yellow bananas and avoids blue bananas by inspecting the score movement. Please follow the Step 6 : Watch the Agent Play.

License

This repository is under the MIT license.