🚦 VisionX: AI-Powered Smart Traffic Management

VisionX is a next-generation traffic optimization system built for Smart India Hackathon 2026. It replaces inefficient fixed-time traffic lights with a Deep Reinforcement Learning (DQN) agent that dynamically manages intersection phases in real-time, reducing total waiting vehicles by over 85%.

---

✨ Key Features

• 🧠 Reinforcement Learning Core: A 64-neuron PyTorch DQN trained via Experience Replay to optimize traffic flow dynamically.
• ⏱️ Dynamic Green Timing: Automatically calculates optimal phase durations (10s – 45s) based on live queue lengths.
• 🚨 Emergency Overrides: Manual UI triggers to instantly clear routes for Ambulances or enforce "Police All-Stop" red lights.
• 🌐 Full-Stack Dashboard: A live React.js dashboard communicating with the AI via a Node.js synchronization bridge.

---

🏗️ System Architecture & Folder Structure

This project uses a hybrid OS architecture. The graphical interface and server run on Windows, while the AI training and simulation physics run on WSL (Ubuntu/Linux) for maximum GPU efficiency.

VisionX-Smart-Traffic/
├── ai_agent/          # PyTorch Model & SUMO Simulation (Run in WSL)
│   ├── rl_agent.py    # Training script
│   └── run_demo.py    # Production script (Used for Pitch)
├── backend/           # Node.js Data Bridge (Run in Windows PowerShell)
│   └── server.js
└── frontend/          # React Dashboard (Run in Windows PowerShell)
    └── src/App.jsx

---

⚙️ Prerequisites

Before running the system, ensure you have the following installed:

Environment	Requirement
Windows	Node.js (v16+)
WSL / Linux	Python 3.10+, PyTorch, NumPy, Requests
WSL / Linux	Eclipse SUMO (Simulation of Urban MObility)

Note: Ensure SUMO_HOME is added to your WSL environment variables after installing SUMO.

---

🚀 Installation & Setup

1. The Backend Bridge (Windows)

Open a Windows PowerShell or Command Prompt terminal.

cd backend
npm install express cors

2. The Frontend Dashboard (Windows)

Open a second Windows PowerShell terminal.

cd frontend
npm install
npm install -D tailwindcss postcss autoprefixer

3. The AI Agent (WSL / Linux)

Open your WSL / Ubuntu terminal.

cd ai_agent
pip install torch numpy requests traci

---

🌐 Critical Network Configuration

⚠️ IMPORTANT: Because the AI agent runs inside WSL and the backend server runs on Windows, you must manually link their IP addresses before running the demo.

1. Run ipconfig in Windows PowerShell and copy your IPv4 Address (e.g., 192.168.1.15).
2. Open ai_agent/run_demo.py and replace the BACKEND_URL variable with your actual IP address:

BACKEND_URL = "http://192.X.X.X:3000"  # Replace with your IPv4

---

🎬 Running the Live Pitch Demo

Launch the three components in this exact order to ensure the backend is ready before the AI begins sending data.

Terminal 1 — Start the Backend Server (Windows)

cd backend
node server.js
# Expected Output: 🚀 Pro Bridge Active on Port 3000

Terminal 2 — Start the Frontend UI (Windows)

cd frontend
npm run dev
# Then open http://localhost:5173 in your browser

Terminal 3 — Launch the AI Simulation (WSL)

cd ai_agent
python3 run_demo.py
# Expected Output: 🚦 Production Agent Live. Listening for React Overrides...

The SUMO GUI will open automatically. The React Dashboard will instantly sync and begin displaying live traffic data.

---

📊 Performance Impact: Fixed-Time vs. AI

Our rigorous TraCI simulation tests prove the superiority of the DQN model:

Metric	Baseline (30s Fixed Cycle)	VisionX AI
Max queue per lane	10+ vehicles	2–3 vehicles
Cumulative wait time	High (cascading jams)	~85% reduction
Adaptability	None	Real-time dynamic

📈 See real_ai_comparison.png in the repository for the full benchmark data plot.

---

🛠️ Tech Stack

Layer	Technology
AI / ML	PyTorch (DQN), TraCI
Simulation	Eclipse SUMO
Backend	Node.js, Express
Frontend	React.js, Tailwind CSS
OS Bridge	WSL2 (Ubuntu) + Windows