Automated part inspection uses cameras, 3D sensors, scales and gauges to check manufactured parts for defects, dimensions and correct assembly without a person doing the checking. The system decides pass or fail for each part, rejects or sorts the failures, and records the result. This article covers what gets inspected, the main inspection methods, how small and large parts are handled differently, and how inspection connects to sorting and counting.
Part inspection is one of the most common uses of machine vision and one of the strongest automation targets, because manual inspection is slow, tiring and inconsistent from shift to shift.
What automated part inspection checks
- Surface defects: scratches, pits, dents, cracks, burrs, stains and contamination.
- Dimensions: lengths, diameters, hole positions, gaps, flatness and profile.
- Presence and assembly: missing features, wrong components, orientation and seating.
- Mass: weight against a tolerance band, which reveals voids, missing material and mixed parts.
- Identity: part type, labels, codes and markings.
Part inspection methods compared
| Method | What it finds | Best for |
|---|---|---|
| 2D machine vision | Surface defects, presence, orientation, codes, 2D dimensions | Most visual checks at high speed |
| 3D laser profiling | Height, warp, flatness, volume, profile | Shape and dimensions that a flat image cannot show |
| Dimensional metrology | Precise measurements against tolerances | High-tolerance parts and critical features |
| Surface defect detection | Fine scratches and cosmetic flaws on reflective or textured parts | Finished and cosmetic surfaces |
| Weighing | Mass outside tolerance | Voids, missing material, mixed parts |
| AI-based image analysis | Variable or cosmetic defects that rules struggle to describe | Natural variation and subtle defect types |
Many systems combine methods. A robotic cell, for example, might weigh each part, scan it in 3D and image every face before deciding pass or fail.
Small parts and large parts
Small and high-tolerance parts
Small parts are usually inspected in a benchtop station or a robotic cell, where the part can be presented precisely to cameras and sensors. Sciotex’s robotic inspection cell uses a robot to pick each part, weigh it, scan it in 3D and image it for surface defect classification and metrology, then sort it by result. See also robotic inspection.
Large parts and parts in motion
Large parts are usually inspected where they already move, on a conveyor, so they never have to be handled for inspection. Synchronized 2D and 3D stations check the top, the underside and the shape as each part passes, as in the ConveyorVIEW conveyor inspection system.
From pass/fail to sorting and counting
An inspection result is only useful if something happens to the part. Automated inspection usually ends in one of three actions:
Combining inspection with sorting and counting removes several manual steps at once, which is why it is one of the highest-return automation projects. For counting by itself, see the automated counting guide.
Data and traceability
Every inspection produces data: pass/fail, measurements, defect type, images and time. Logged per part or per lot, it gives quality teams traceability, shows trends before they become scrap, and points to the upstream process causing defects.
Choosing a part inspection system
- Define the defects and dimensions that matter, with examples of good and bad parts
- Set the rate, accuracy and acceptable false reject rate
- Decide what happens to failed parts: reject, sort or hold
- Decide what data you need and where it should go
- Test the approach on real production samples before buying
Part variation decides what works, so testing real parts is the most important step. Sciotex tests customers’ sample parts free of charge to confirm the right configuration. Compare options on our automated inspection systems page, or browse inspection systems Sciotex has built.
Frequently asked questions
It is the use of cameras, 3D sensors, scales and gauges to check manufactured parts for defects, dimensions and correct assembly automatically, with a pass or fail decision and a record for every part.
Common methods are 2D machine vision, 3D laser profiling, dimensional metrology, surface defect detection, weighing and AI-based image analysis. Many systems combine several.
Usually on the conveyor they already travel on, with synchronized cameras and 3D sensors checking the top, underside and shape as each part passes, so the part does not need to be handled.
Yes. Failed parts can be rejected or sorted by defect class, and good parts counted into packaging in the same system, removing several manual steps at once.
Related articles
- Machine Vision: The Complete Guide to Machine Vision Systems
- Robotic Inspection: How It Works, Types and Where It Fits
- Automated Dimensional Inspection: Methods, Accuracy and How to Choose
- Scratch Detection: How to Find Surface Defects on Reflective Parts
- Industrial Automation: A Practical Guide to Manufacturing Automation