Introduction

This project enables automated quality inspection of helical gears manufactured by Sona BLW Precision Forgings Ltd (Sona Comstar) for Tesla. It integrates a rolling machine, Baumer industrial cameras, and a YOLO-based object detection system to ensure defect-free production.

System Workflow

Step 1: Gear Rotation & Image Capture Setup

The manufactured helical gear is placed in the rolling machine. The machine uses a drive mechanism to rotate the gear 360° based on a PLC signal.

Step 2: Image Acquisition

Two Baumer industrial cameras capture images during rotation. A total of 30 images are acquired within 1 minute for complete surface inspection.

Step 3: Image Preprocessing and perspective transformation

The images are sent to a FastAPI server (running on Uvicorn). Using OpenCV, a perspective transformation is applied to convert tilted images into a 2D planar view for accurate object detection.

Step 4: Defect Detection Using YOLO

Defect Classes: 1. Handling dent 2. Root grinding 3. ChamferMiss 4. step grinding 5. Flank_unclean 6. Rust 7. Heat treatment dent.

The transformed images are sent to the YOLO object detection model. The model analyzes the images and detects defects .

Step 5: Data Storage

The inspection results are saved in the Sona Comstar database server for tracking and analysis.

Step 6: Real-Time Monitoring

Simultaneously, results are displayed on a real-time dashboard over the local LAN IP. Operators can monitor inspection status and detected defects instantly.

Technologies Used

Hardware: Rolling Machine, PLC, Baumer Industrial Cameras

Software: FastAPI, Uvicorn, OpenCV, YOLO Object Detection

Database: Sona Comstar Database Server

Dashboard: Real-time visualization of Production Dashboard over Local LAN