UC-15 · Robotic · arc · spot

Weld Seam Inspection

A weld that looks fine to the eye can hide porosity, undercut, or insufficient penetration. Post-weld vision inspection — combined with thermal imaging where available — catches the defects that ultrasonic testing only finds at the QA station, hours after the weld cooled.

REC INSPXION X-12B LIVE INSPXION X-12B · WELD SEAM WELDS/HR · 480 0.6% MISS LIVE InspXion SPEC CLASSES 5 defects 3D PROFILE LASER LINE CYCLE REAL-TIME OEM FANUC · ABB PASS · BEAD OK WELDS 8,142 REJECTS 47 REWORK 3
Why it matters

Weld defects discovered at QA mean the entire fixture rebuilds.

When a defect is found at end-of-line QA, the part is scrap and the fixture upstream may need adjustment. Inline weld inspection catches the defect in the seconds after the weld is laid — before the next weld, before the part advances. The fix is a parameter adjustment, not a rebuild.

Two paths to the same outcome

Retrofit, or build it in from the start.

Retrofit · use what you have

If you have a robotic welder with no inline verification.

We add a weld-following camera to the existing robot end-of-arm tooling. The camera images each weld immediately after the torch passes; AI grades the weld for porosity, undercut, spatter, and continuity. Bad welds trigger an immediate alert to the operator; the robot pauses before the next part is loaded.

  • Mount on existing robot EOAT — no cell modification
  • Weld-following: imaged in the seconds after torch passes
  • Defect classes: porosity · undercut · spatter · burn-through · missing
  • Real-time alert to operator + robot pause
  • Per-weld telemetry to MES for traceability
Greenfield · spec from scratch

If you're designing a new welding cell.

Spec the IX-AS/01 with a heat-resistant enclosure and laser line projector for 3D weld profiling. The combination resolves porosity, undercut depth, and bead height — the three measurements that predict weld quality. Per-weld results feed the welding controller for closed-loop parameter adjustment.

  • Heat-resistant enclosure for cell-mount installation
  • Laser line profiling: porosity · undercut depth · bead height
  • Closed-loop welder integration: auto-tune voltage, current, travel speed
  • Per-weld 3D profile + image archived for traceability
  • Pre-trained on MIG, TIG, spot, and laser-weld signatures
How it works

Four steps from camera to verdict.

01

Follow

Camera follows the torch. Each weld imaged within 1-2 seconds of laydown.

02

Profile

2D image + (optional) 3D laser-line profile capture surface and cross-section.

03

Grade

AI evaluates each weld against quality classes. Per-class severity score.

04

Act

Bad weld → operator alert + robot pause. Good weld → next part loaded. Telemetry logged.

Compatible with

The InspXion stack for this use case.

Ready to put this into production?

Get a sized configuration in 48 hours.

Tell us your line, your existing equipment, and your tolerance window. We'll come back with a retrofit assessment, a greenfield specification, and a rough cost estimate within two business days.