PathfindingDeepQLearn

Program aims to solve the problem of global pathfinding in mobile robots by making use of Deep Q-Learning.

A DQN model is defined, and then is run on a procedurally generated random environment. The agent has to navigate through the environment from a start state to the goal state. The output is shown in the form of a search graph,each representation showing every subsequent step taken by the agent. This project solves self-made maze in a variety of ways: A-star, Q-learning and Deep Q-network.

Directions to Use

The Code is present in a iPython Notebook. Install jupyter and open the notebook.

Python library Dependencies are as follows:

1. Numpy
2. Pandas
3. TensorFlow
4. Keras These dependencies can be resolved by installing them using pip command.

Remember to use Python 3.5 with Tensorflow.