saadmahboob / rf_helicopter Goto Github PK

The intention of this project is to implement a grid world representation of the Helicopter Game and to train an Agent to learn to play the Game successful.

The Agent should be able to generalise well when navigating any course.

In additional to the obstacles as seen in the original game additional complexity has been added in the form of wind.

The wind component means that the Agent needs to be able to not only navigate the course but also manage the complexities of the wind.

This project has been implemented in Python, utilising Numpy, Keras and Tensorflow.

For full documentation please see the MKDocs site located in the /docs directory, instructions have been added to the README.md file.