001 · Automotive · Certainty AI

Automotive Glass Inspection Automation

Clarity on every road. Precision inspection for windshields, sidelites, backlites, and more with Certainty AI.

Book a Call Contact Us
LIVE · Glass Cell
VISION AI
Automotive glass inspection
Scratch
10-20µm
Cycle
30-120s
Accuracy
>99%
Status
PASS
002 · Challenges & Solutions

Why automotive glass is challenging — and how we solve it

Automotive glass must meet demanding optical and safety requirements while integrating various functional components.

Challenges

  • Transparency & glare: highly polished surfaces make scratches and sleeks easy to miss
  • Curvature & thickness: bending and laminating introduce refraction and double images
  • Functional layers: ceramic frit, coatings, HUD zones, conductive heaters, encapsulation
  • Edge quality: long perimeters with holes/slots increase risk of chips and micro-cracks

Intelgic Solutions

  • Lighting matched to panel and task to reveal defects without glare
  • Custom machine vision design tuned for each glass model and curvature
  • Certainty AI trained on automotive datasets to separate true defects
  • Microscopic capability detects ≈10-20 µm scratches and edge chips ≥20 µm
003 · Intelgic Approach

Turnkey solution combining optics, motion, AI

Application-specific lighting geometry, multi-axis motion, and Certainty AI engineered for automotive transparencies.

Advanced Lighting

Low-angle oblique + controlled dark-field reveals hairline scratches without glare. Telecentric backlight creates crisp silhouettes for edge inspection.

Certainty AI

Trained on automotive datasets to separate true defects from heater lines, prints, and structured reflections with high accuracy.

Digital Quality Certificate

Tamper-evident digital record for every PASS panel with defect snapshots, measurement summaries, and full traceability.

004 · What We Inspect

Comprehensive defect detection

Across all critical areas of automotive glass.

Surface & Cosmetic

Hairline scratches, sleeks, scuffs, digs/pits, stains, water spots, haze and rainbow artifacts on coated zones.

Frit & Printing

Ceramic frit pinholes, voids, smears, edge raggedness. Logo/mark presence, contrast, size, and position.

Edges & Cutouts

Perimeter chips, micro-cracks, shelling; bevel/chamfer quality. Hole/slot presence and size; VIN viewport sizing.

Functional Features

ADAS camera & sensor brackets position/orientation. Backlite defogger lines continuity. Encapsulation quality.

Geometry & Optical

Outer contour vs. CAD overlay. Curvature/sag checks. Optional deflectometry for distortion/slope mapping in HUD region.

Traceability

OCR/OCV of part numbers, E-marks, date/shift codes; barcode/2D code verification for complete traceability.

005 · System Architecture

Precision components in harmony

Engineered for reliable inspection of automotive transparencies.

Imaging Hardware

  • High-resolution area-scan (12-65 MP) with HDR for reflective surfaces
  • Telecentric lenses for metrology; high-NA macro for micro-defects
  • Multiple lighting geometries matched to inspection tasks

Motion & Control

  • Multi-axis rails/gantry for full-field coverage
  • Rotary/tilt stages for angle control
  • Soft-nest/vacuum fixtures to protect surfaces

Software & Integration

  • Certainty AI with reflection-aware segmentation
  • Integration with PLC/robot, MES/ERP via OPC-UA, Modbus/TCP
  • Dashboard & KPIs for yield analysis and defect trends
006 · Technical Specifications

Performance & hardware specs

Performance & Detection

  • › Surface scratch/pit detection: ≈10-20 µm on flats; ≈20-30 µm on curves
  • › Edge chips: ≥20 µm length; bevel/chamfer measurement to ±0.05-0.10 mm
  • › Bracket position: ±0.2-0.5 mm and ±0.2-0.5° typical
  • › Cycle time: 30-120 seconds per panel for full checks
  • › Defect classification accuracy: >99% for critical defects

Hardware Components

  • › Cameras: 12-65 MP area; optional 2k-8k line-scan; HDR/global shutter
  • › Optics: telecentric 0.3-1.0×; macro up to 5×; liquid lens autofocus
  • › Lighting: low-angle oblique, collimated backlights, coaxial/dome, polarization
  • › Motion: multi-axis rails/gantry; rotary/tilt stages; encoder sync
  • › Optional deflectometry for slope/distortion mapping
007 · FAQ

Frequently asked questions

Yes. AI models are trained on heater line patterns and use frequency/shape cues to avoid false positives; low-angle dark-field highlights true micro-scratches.
Yes. We measure bracket pose vs. datums and check bead presence, width, and squeeze-out. Orientation errors and missing/extra components are flagged.
Optionally, a deflectometry module maps slope/distortion and wedge proxies in the driver eye-box and HUD region for process tuning and acceptance decisions.
With multi-axis rails/gantry and soft-nest or vacuum fixtures that stabilize the part while keeping optical surfaces safe.
Per-panel DQC, shift/day reports, defect Pareto, and full audit logs with user actions; secure sharing for OEM/Tier-1 stakeholders.

Solution

Industrial AI

LEGAL

Head Quarter

17352 Murphy Ave Suite 101
Irvine, CA 92614
Phone: (949) 317-2420

Branch Office

5ES7-H, 5Th Floor, East Tower, Mani Casadona, Plot no -IIF/04, Action Area -IIF, Newtown, Kolkata-700160

Certificate

intelgic ISO 27001 certification

Social links

©2026 Intelgic Inc. All Rights Reserved.