QR on Track: AI-Enabled TrackFit Scanner

Hinglish Name: Railfit BarCode Scanner
Purpose: A smart system for Indian Railways to mark QR codes on track fittings (metal clips, rubber pads, concrete sleepers) and track them using an AI-powered mobile app with augmented reality (AR). It connects to railway portals (UDM and TMS) to monitor quality, manage inventory, and ensure safety. Built for Smart India Hackathon 2025.

Team: 6 members (2 Hardware, 2 Machine Learning, 2 Full-Stack).

What We’re Building

Hardware: 2D Laser Marking System

What It Does: A small machine that etches tiny QR codes (5–10mm) on railway fittings to store info like vendor, supply date, and inspections.

Parts:

• ESP32 Chip (~₹400–800): Controls the system and uses Wi-Fi to send data.
• Two Nema-17 Motors (~₹1,200 each): Moves the laser to draw QR codes.
• Engraver-2 Laser (10W, ~₹8,000–16,000): Etches codes on metal, rubber, and concrete.
• Aluminum Frame (20x20cm, ~₹2,000): Holds everything together.
• Camera (OV2640) (~₹400): Checks if the fitting is metal, rubber, or concrete to adjust the laser.
• Battery (12V) (~₹2,000): Makes it portable for field tests.
• Enclosure (~₹1,000): Keeps the laser safe and dust/water-proof.

How It Works:

• The ESP32 moves the laser to etch QR codes on fittings, guided by the motors.
• The camera tells the system what material it’s marking, and AI adjusts the laser for clear codes.
• A feeder moves fittings under the laser (1–2 seconds per code).
• Wi-Fi sends marking logs to a database, and the battery lets it work anywhere.

Why It’s Great:

• Works on different materials without manual tweaks.
• Cheap (~₹20,000–30,000) and portable for railway depots.
• Safe and tough for outdoor use.

Software: AI-Powered QR Scanner and Management

What It Does: A phone app and website that scan QR codes, show fitting details, predict problems with AI, and use AR to help workers in the field.

Parts:

1. Website:
   • Scans QR codes with a phone camera, showing vendor, supply date, warranty, and inspection info.
   • Works offline and supports English/Hindi.
   • Connects to UDM (www.ireps.gov.in) and TMS (www.irecept.gov.in) for data.
2. AI Engine (Python):
   • Predicts faulty fittings (using XGBoost).
   • Spots unusual issues (DBSCAN) and predicts low stock (ARIMA).
   • Learns from worker inputs (e.g., “clip is rusted”) to get smarter.
3. AR Interface (ARCore/ARKit):
   • Shows fitting info as 3D overlays on a phone or AR glasses (e.g., red flag for faulty parts).
   • Guides workers with arrows to fix issues.
   • Takes voice commands (BERT) like “Show defective clips.”
4. Web Dashboard (Django/React):
   • Shows graphs of quality trends and inventory.
   • Sends email/SMS alerts for problems (e.g., low stock).
   • Securely syncs with UDM/TMS.

How It Works:

• Workers scan a fitting’s QR code to see its details.
• AI checks for issues (e.g., bad batches) and predicts stock needs.
• AR shows info right on the fitting, and voice commands make it hands-free.
• The dashboard helps managers see trends and get alerts.
• Worker feedback improves the AI over time.

Why It’s Awesome:

• AR and voice commands make fieldwork easy and modern.
• AI prevents accidents by catching problems early.
• Works on cheap phones and scales to millions of fittings.

Deliverables

• Hardware: A laser system marking QR codes on 500 sample fittings.
• Software: A Website with QR scanning, AR, and AI, plus a web dashboard.
• Integration: Mock connection to UDM/TMS for data and alerts.

Team Roles

Hardware Engineers

• Lead Hardware Developer:
  • Build/program the ESP32, motors, and laser for QR marking.
  • Test on sample fittings (metal, rubber, concrete).
• Hardware Testing and Safety:
  • Add camera, battery, and safe enclosure.
  • Test QR codes for readability and durability.
• Tools: Soldering kit, multimeter, Arduino IDE, 3D printer (optional).

Machine Learning Engineers

• AI Model Developer:
  • Build AI to detect materials and predict quality/inventory issues.
  • Deploy models on cloud (AWS/GCP).
• NLP and Feedback Specialist:
  • Add voice commands and AI feedback loop.
  • Optimize for edge devices (e.g., Raspberry Pi).
• Tools: Python, TensorFlow, OpenCV, Hugging Face, Colab.

Full-Stack Developers

• Frontend Developer:
  • Build React app with QR scanning and AR.
  • Add offline storage and English/Hindi UI.
• Backend and Dashboard Developer:
  • Create backend and React dashboard.
  • Mock UDM/TMS APIs and add alerts.
• Tools: Django, React, Postman, VS Code.

Feasibility

• Cost: ~₹20,000–30,000 (ESP32: ₹800, Nema-17: ₹2,400, Engraver-2: ₹10,000, frame/camera/battery: ₹7,000).
• Resources: Buy from Robu.in or Amazon India. Use free tools (Flutter, Python) and datasets (Roboflow QR, Kaggle Barcode, Rail-5k).
• Scalability: Wi-Fi and cloud make it work for millions of fittings.
• Hackathon Fit: Buildable in 3–4 months, innovative with AR/AI.