JoshVarty/BananaCollector_DoubleQLearning

An implementation of DoubleQLearning to solve Unity's Banana Environment

Banana Collector

Unity's Banana Collector Environment is an environment in which an agent must collect as many yellow bananas (+1) as possible while avoiding blue bananas (-1).

The agent interacts with the environment via the following:

• It is fed observations of the current state via a vector of 37 elements
• It can choose to make any of 4 actions: (Left, Forward, Right or Back)

Sample image taken from: https://github.com/udacity/deep-reinforcement-learning/tree/master/p1_navigation

This repository trains an agent to attain an average score (over 100 episodes) of at least 13. It trains the agent using the Double DQN Reinforcement Learning algorithm.

Prerequisites

• Anaconda

• Python 3.6

• A conda environment created as follows

  • Linux or Mac:

  conda create --name drlnd python=3.6
  source activate drlnd 
  

  • Windows

  conda create --name drlnd python=3.6 
  activate drlnd
  

• Required dependencies

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

Getting Started

1. git clone https://github.com/JoshVarty/BananaCollector_DoubleQLearning.git

2. cd BananaCollector_DoubleQLearning

3. Download Unity Banana Collector Enviroment:

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

4. Unzip to git directory

5. jupyter notebook

6. You can train your own agent via DQN.ipynb or watch a single episode of the pre-trained network via Visualization.ipynb

Results

In my experience the agent achieves an average score of 13 after ~400 episodes of training:

A sample run generated from Visualization.ipynb

Notes

• Only tested on Ubuntu 18.04
• Details of the learning algorithm and chosen architecture may be found in Report.md