Smart Parking RL - Deep Reinforcement Learning for Intelligent Parking

🚗 Overview

Smart Parking RL is a deep reinforcement learning project that simulates an intelligent parking agent navigating a grid environment with obstacles and designated parking spots. The agent is trained using Deep Q-Learning (DQN) to find the shortest and safest path to park optimally.

---

👤 Author

Rashin Gholijani Farahani
MSc Computer Engineering Applicant - Passionate about AI & Mobility Solutions

---

🔍 Project Objectives

• Design a realistic, grid-based parking environment
• Implement a Deep Q-Network (DQN) to train a car agent
• Reward the agent for reaching valid parking spots
• Penalize for collisions or invalid parking
• Compare DQN vs BFS baseline path planning
• Visualize agent behavior via pygame and matplotlib

---

🧠 Core Features

• ✅ Valid parking spot detection
• ⛔ Obstacle-aware navigation
• 🧠 DQN-based learning using PyTorch
• 🔍 BFS path as baseline strategy
• 🎮 Real-time animation with pygame
• 📊 Grid-path visualization with matplotlib

---

📁 Project Structure

SmartParkingRL/
├── dqn_training.ipynb          # RL agent training (PyTorch DQN)
├── smart_parking_env.ipynb     # Custom parking environment
├── Smart Parking RL.ipynb      # Integrated final notebook

---

🛠️ Dependencies

• Python 3.9+
• NumPy
• PyTorch
• Matplotlib
• Pygame

📊 Result Example

• Green = Destination Parking
• Yellow = Valid Slots
• Red = Obstacles
• Blue = Agent Path

🤝 Connect

• 📫 Email: farahanirashin@gamil.com
• 🌐 LinkedIn: https://www.linkedin.com/in/rashin-gholijani-farahani/