/p2_continuous-control

2nd Project of Udacity Deep Reinforcement Learning NanoDegree

Primary LanguageJupyter Notebook

Project 2: Continuous Control

Introduction

For this project, you will work with the Reacher environment.

Trained Agent

In this environment, a double-jointed arm can move to target locations. A reward of +0.1 is provided for each step that the agent's hand is in the goal location. Thus, the goal of your agent is to maintain its position at the target location for as many time steps as possible.

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.

Training the agent

For this project, we will provide you with a version that contains 20 identical agents, each with its own copy of the environment.

We will implement the Continuous Control with Deep Reinforcement Learning paper (Lillicrap et al), that present a model-free, off-policy actor-critic algorithm using deep function approximators that can learn policies in high-dimensional, continuous action spaces known as Deep Deterministic Policy Gradients (DDPG)

The Reacher environment with 20 agents version is specially 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 and we encourage to give it atry with those afterwards.

Solving the Environment

The barrier for solving this version of the environment is 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

To run this project you will need Python 3.6. By using Conda you can create and activate an environment named drlnd, and install all required dependencies:

  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 this repository and go to p2_continuous-control/ folder. If you haven't already, install pip, and finally install all project dependencies.

git clone https://github.com/aamonten/p2_continuous-control.git
cd p2_continuous-control
pip install -r requirements.txt
  1. Create an IPython kernel for the drlnd environment.
python -m ipykernel install --user --name drlnd --display-name "drlnd"

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

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

    (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.

  3. Place the file in the DRLND GitHub repository, in the p2_continuous-control/ folder, and unzip (or decompress) the file.

Instructions

Follow the instructions in Continuous_Control.ipynb to get started with training the agent!