Robot Control with Deep Q-Network

This project demonstrates a robot controlled by a Deep Q-Network (DQN) for learning locomotion with respect to a given sensor's readings. The code uses a simulated environment provided by Webots.

Project Structure

• controllers/drive_robot/

  • drive_robot.py: Main controller script that implements the DQN agent and training loop.
  • robot_desc.py: Contains the definition of the robot class with sensor and actuator configurations.

• worlds/

  • world.wbt: Webots world file where the simulation environment is defined.
  • .world.jpg: An image possibly showing the world setup.

Setup Instructions

Prerequisites

• Python 3.x
• Webots (Simulation environment):
  • Download and install Webots from official Webots website.

Installation

1. Install Required Python Packages:

   pip install numpy torch pandas