To train a Actor Critic Policy Gradient Network agent for the Robotic Hand movement in UnityML agents

You can view the publication from DeepMind here

In this environment, a double-jointed arm can move to target locations.

Reward: A reward of +0.1 is provided for each step that the agent's hand is in the goal location.

Goal: The goal of your agent is to maintain its position at the target location for as many time steps as possible.

Observation Space: 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.

Action Space: Every entry in the action vector should be a number between -1 and 1.

In order to consider the environment has been solved, the agent must get an average score of +30 over 100 consecutive episodes.

Clone the repository -

git clone https://github.com/abhisheksgumadi/continuous-control.git .

Install jupyter notebook with the command

pip install jupyter

Then, open the Continuous_Control.ipynb notebook

jupyter notebook Continuous_Control.ipynb

Download the Robotic Hand environment for Unity for MacOSX at here

The code consists of the following modules

Continuous_Control.ipynb - the main notebook
agent.py - defines the Agent that is being trained
model.py - defines the PyTorch model for the Actor and the Critic network
checkpoint_actor.pth - is the final trained Actor network
checkpoint_critic.pth - is the final trained Critic network

Please follow the code in Navigation.ipynb to train the agent

The average reward collection over 100 episodes plotted in a graph below. It shows the reward on the Y axis for every episode on the X axis