Self-Driving Car Simulation

A self-driving car simulation that employs neural network-based techniques, particularly reinforcement learning and deep Q-learning. The neural network demonstrates proficiency in obstacle avoidance and optimal travel time estimation between destinations. The project showcases expertise in autonomous systems and artificial intelligence.

Features

• Neural network-based self-driving car simulation.
• Obstacle avoidance using reinforcement learning and deep Q-learning.
• Optimal travel time estimation between destinations.
• Ability to draw obstacles for the car to avoid.
• Saving and reloading neural network models for specific obstacle courses.

Files

• map.py: The main script containing the self-driving car simulation logic, obstacle handling, and interaction with the neural network AI.
• car.kv: The Kivy language file for defining the graphical user interface (GUI) layout.
• ai.py: Script containing the implementation of the neural network architecture, experience replay, and deep Q-learning algorithm.

Installation

Clone the repository:

git clone https://github.com/amirzarandi/self-driving-car
cd self-driving-car

Usage

Run the simulation:

Copy code python map.py The GUI window will open, displaying the self-driving car simulation. Use the left mouse button to draw obstacles on the map. Observe the self-driving car's behavior in avoiding obstacles and reaching its destination. Documentation

Detailed documentation and explanations of the neural network architecture, reinforcement learning techniques, and deep Q-learning algorithm can be found in the comments within the code files (map.py and ai.py).