Navigation

Introduction

For this project, you will train an agent to navigate (and collect bananas!) in a large, square world.

A reward of +1 is provided for collecting a yellow banana, and a reward of -1 is provided for collecting a blue banana. Thus, the goal of your agent is to collect as many yellow bananas as possible while avoiding blue bananas.

The state space has 37 dimensions and contains the agent's velocity, along with ray-based perception of objects around agent's forward direction. Given this information, the agent has to learn how to best select actions. Four discrete actions are available, corresponding to:

• 0 - move forward.
• 1 - move backward.
• 2 - turn left.
• 3 - turn right.

The task is episodic, and in order to solve the environment, your agent must get an average score of +13 over 100 consecutive episodes.

Getting Started

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

   • 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 to obtain the environment.

2. Place the file in the DRLND GitHub repository, in the p1_navigation/ folder, and unzip (or decompress) the file.

Instructions

Follow the instructions in Navigation.ipynb to get started with training your own agent!

(Optional) Challenge: Learning from Pixels

After you have successfully completed the project, if you're looking for an additional challenge, you have come to the right place! In the project, your agent learned from information such as its velocity, along with ray-based perception of objects around its forward direction. A more challenging task would be to learn directly from pixels!

To solve this harder task, you'll need to download a new Unity environment. This environment is almost identical to the project environment, where the only difference is that the state is an 84 x 84 RGB image, corresponding to the agent's first-person view.

You need only select the environment that matches your operating system:

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

Then, place the file in the p1_navigation/ folder in the DRLND GitHub repository, and unzip (or decompress) the file. Next, open Navigation_Pixels.ipynb and follow the instructions to learn how to use the Python API to control the agent.

(For AWS) If you'd like to train the agent on AWS, you must follow the instructions to set up X Server, and then download the environment for the Linux operating system above.

Installation instuctions

ML-Agent - https://github.com/Unity-Technologies/ml-agents/blob/master/docs/Installation-Windows.md

Anaconda/TensorflowGPU - https://github.com/Unity-Technologies/ml-agents/blob/master/docs/Installation-Windows.md

Implementation details

The code has 2 primary parts

1. Train-Test
2. Class declarations

The Train-test code is in the Navigation.ipynb notebook Class declarations are in model.py and dqn_agent.py

Classes used:

QNetwork - Neural Network declaration Agent - Agent class to perform steps and learn ReplayBuffer - Buffer to store State-Action-Reward-NextState tuples to sample training data from

The implementation uses Fixed Q values (2 Neural networks) and Replay buffer methods explained in Deep Q Learning paper.

Training

• The training stage implemented performs typical Deep-Q learning training steps.
• The model weights are saved in checkpoint.pth file (When average score is >14)
• The training runs for 2000 epochs by default, with 1000 max number of steps per epoch
• There is decaying epsilon to achieve gradually decreasing randomness in actions taken

Test

• The test phase loads the latest model from file system and getting score for the tests performed.

Useful links

Prioritized Experience Replay - https://github.com/rlcode/per
Deep learning frameworks comparison - https://www.netguru.com/blog/deep-learning-frameworks-comparison
Rainbow paper (DQN paper methods comparison) - https://arxiv.org/pdf/1710.02298.pdf
OpenAI Retro (Rainbow all menthods) - https://medium.com/intelligentunit/conquering-openai-retro-contest-2-demystifying-rainbow-baseline-9d8dd258e74b

How is value based DQN different from policy based DRL

1. In Value based NN, the output neurons are activated. Then we use Epsilon greedy policy to select the neuron (action) with highest value
2. In policy base NN implementations, the actication of neurons is the probability for that action! No more epsilon greedy approach, instead use these probabilities for selecting the action