This project utilizes the Duckietown platform to develop a lane-following autonomous robot using reinforcement learning. Specifically, it employs the Q-learning algorithm, a popular reinforcement learning approach, for navigating a Duckiebot through a Duckietown environment. The project is designed for educational and research applications, providing a scalable platform for understanding complex tasks in perception, decision-making, and autonomous control.

The Gazebo simulation environment includes:

• A 3D representation of Duckietown with roads, traffic lights, and various objects.
• An interface for controlling the Duckiebot through predefined actions like "Turn Left," "Turn Right," and "Forward."

The Duckiebot is a custom-built robot equipped with:

• Camera: Used for lane detection.
• Motors: Controlled via an H-bridge motor controller.
• Raspberry Pi: Runs the Q-learning algorithm and communicates with the sensors and actuators.

1. Clone the repository:

   git clone <repository_url>

2. Install the required dependencies:

   sudo apt update
   sudo apt install ros-noetic-desktop-full gazebo python3-opencv python3-numpy

3. Source the ROS environment and build the workspace:

 source /opt/ros/noetic/setup.bash
 catkin_make
 source devel/setup.bash

1. Launch the simulation with one mobile robot:

   roslaunch duckietown_gazebo main.launch

2. Run the Q-learning training module:

   roslaunch lane_following start_training.launch #this run start_qlearn_lane_follow.py for tranning the agent which saved A dedicated checkpoint file.

This project is based on resources and tools from Duckietown and the educational platform The Construct, where we developed and tested the Duckiebot's Q-learning functionality in simulation. Special thanks to The Construct for providing a comprehensive learning environment.