This project is part of Udacity Deep Reinforcement Learning Nanodegree. This project aims to develop and train a Deep Reinforcement Learning (RL) agent to navigate and collect bananas in one of Unity environments.
The goal of the agent is to navigate in the environment to collect as many yellow bananas as possible while avoiding blue bananas. A reward of +1 is given for collecting a yellow banana, and -1 for a blue banana.
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 the agent is considered successful in solving the environment when it gets an average score of
- Dependencies. Set up the python environment by following these steps.
- Unity Environment. Set up the Unity environment by following these steps.
- Deep Reinforcement Learning Agent. A Double Deep Q-Network (Double DQN) agent is used to solve the environment in this project. The agent is defined in agent.py file, and each of the two Deep Q-Network is defined in model.py. The Navigation.ipynb notebook contains a script for training the agent and demonstrates that the agent can successfully solve the environment within a few hundred episodes.