Visual Disc Screener: Revolutionizing Small-Part Appearance Inspection
Project Background
A bearing manufacturer faced the pain points of low efficiency and poor consistency in manual visual inspection. Operators could inspect only 30–60 pieces per minute, and fatigue led to persistently high miss rates. The customer needed an automated solution capable of high-speed, high-precision, 100% appearance defect inspection.
Solution
AIMA ROBOT designed a deep-learning-based visual disc screener that integrates vibratory feeding, disc alignment, multi-angle imaging, real-time AI classification, and automatic sorting into a single machine.
Key Specifications
| Parameter | Value |
|---|---|
| Inspection Speed | 300–600 pcs/min |
| Inspection Accuracy | ≥0.02mm |
| Inspection Items | Scratches, cracks, missing material, mixed parts, dimensions |
| Camera Configuration | 4 industrial cameras, synchronized multi-angle |
| Algorithm | Deep Learning CNN + traditional edge detection fusion |
| Sorting Outputs | OK / NG dual channels, automatic counting |
Performance Comparison
| Metric | Manual Inspection | Visual Disc Screener |
|---|---|---|
| Speed | 30–60 pcs/min | 300–600 pcs/min |
| Accuracy | ≥0.1mm | ≥0.02mm |
| Consistency | Operator-dependent | 99.5%+ |
| Manpower | 6 operators/shift | 1 operator/shift |
Customer Benefits
✅ Good-part detection rate improved to 99.5% · ✅ 83% reduction in labor · ✅ ROI achieved in 6 months · ✅ Traceable data enables continuous quality improvement
