Vision-Guided Mechatronic Gantry System

AI-guided crane automation with perception and edge deployment

I designed and automated a 3-DOF mechatronic gantry system for industrial steel coil loading and unloading. The project combined mechanical design, computer vision, and embedded systems to solve high-impact bottleneck issues in current facility operations.

Loading: Automated Gantry Crane
Unloading: Automated Gantry Crane

Key work

  • Designed and assembled a 3-DOF mechatronic crane and custom C-Hook lifting mechanism.
  • Programmed a spatial perception pipeline using Canny edge detection and contour extraction to calculate payload coordinates and orientation.
  • Fine-tuned a real-time YOLOv8 model for targeted object and wagon detection.
  • Deployed a custom web interface and API on an Arduino UNO WiFi server to synchronize AI perception with real-time stepper motor actuation.

What this project demonstrates

It brought together mechanical engineering, robotics perception, embedded systems, and intelligent automation into a single integrated workflow.

Source code: Arduino UNO ESP32