Present
Feeder tooling and the outfeed create a controlled face, height, position and gap.
Orient · inspect · reject · record
Feeder vision inspection systems.
Verify visible orientation and quality features after centrifugal feeding, then pass, recirculate or reject each component through controlled line logic.
Direct answer
What does vision add to a parts feeder?
Machine vision checks the image of each presented component against defined rules. In a centrifugal feeding system it can confirm orientation, identify visible variants or defects and command a reject or recirculation device before the part reaches the production hand-off.
The camera works only as well as its presentation. Parts must be separated, stable, in focus and illuminated so the required feature is consistently visible across normal production variation.
- Orientation verification before release
- Visible feature, colour or assembly checks
- Controlled reject or recirculation
- Pass, fail and fault signals to the PLC
- Optional counting, images and result records
Inspection architecture
Build the image condition before choosing the camera.
A stable inspection result depends on part presentation, optics, lighting, decision logic and physical rejection as one engineered chain.
Illuminate
Backlight, ring, diffuse or directional lighting makes the relevant feature repeatably visible.
Capture
Lens, working distance, exposure and trigger timing freeze the component at useful resolution.
Classify
Rules distinguish accepted orientation, known variants, rejects and uncertain results.
Act
The PLC tracks the part to an air jet, diverter, stop gate or recirculation point.
Verify
Reject confirmation, blocked-path sensing and counts show that the physical action occurred.
Common inspection goals
Define one measurable question at a time.
“Check the part” is not a specification. Each feature needs an accepted state, a worst-case sample and a response when the result is uncertain.
Top or bottom
Identify which face is visible when two mechanically stable states otherwise appear similar.
Leading end
Confirm a slot, flange, thread, opening or other end feature is leading correctly.
Colour or variant
Identify an approved colour or distinguish visible format features, subject to controlled lighting.
Assembly presence
Check that a defined insert, ring, seal or visible component is present in the field of view.
Mark or print
Inspect presence or position of a visible code, symbol or print where resolution allows.
Gross defect
Detect a defined visible shape or surface failure using approved good and reject boundaries.
Mechanical orientation plus vision verification
The most robust high-speed route often uses mechanical tooling to remove obvious wrong states and vision to verify the final visible feature. This reduces image complexity, protects inspection time and prevents the camera becoming responsible for uncontrolled overlap or random presentation.
Plan for uncertain results
A real image can be affected by contamination, reflections, component variation, motion blur or partial obstruction. The system should distinguish a confirmed pass from fail and no-read conditions. The customer must decide whether a no-read is rejected, recirculated or stops production.
Reject tracking is part of inspection
When the camera and reject device are separated, the PLC must track the inspected component reliably. The reject path should prevent bounce-back and include confirmation where a missed reject would create product or process risk. A reject-bin full sensor, lockable access or reconciliation may also be required.
Use representative good and reject samples
Commissioning should include every approved format, normal colour and finish variation, borderline conditions and known defects. Artificial samples can help develop a concept, but final acceptance rules should reflect actual production parts and the customer’s quality decision.
Inspection data and traceability
Basic systems may provide only pass, fail and fault signals. More detailed scopes can retain counts, reject categories, images, recipe versions or batch references. Data ownership, retention, user access, time synchronisation and higher-level interfaces need explicit definition.
Important boundary
Machine vision can verify only features visible with the agreed view, lighting, resolution and component presentation. It should not be described as checking hidden or unspecified defects.
Related planning pages
Feeder vision FAQ
Questions that define a credible inspection.
Why add machine vision to a centrifugal feeder?
Vision can verify a visible orientation feature that mechanical tooling cannot distinguish reliably, inspect defined attributes and stop or reject an incorrect part before it reaches the next process.
Does a camera replace feeder tooling?
Usually not. Parts still need controlled separation, spacing and presentation so the camera receives a consistent image. Mechanical pre-orientation often reduces the number of states the vision system must handle.
What can feeder vision inspection check?
Depending on image view, lighting, resolution and component presentation, it may check orientation, presence, colour, printed marks, shape, assembly features or defined surface conditions. Capability must be proved with representative good and reject samples.
What happens to a failed component?
A failed part may be blown off, diverted, recirculated or stopped at a controlled reject station. The design should prevent rejected parts re-entering accepted output unless that route is intentionally validated.
Can inspection results be recorded?
Yes, where specified. The control scope can include counts, reject categories, images, batch information or interfaces to higher-level systems. Retention and traceability requirements must be agreed before design.
Technical content reviewed by Sortation Solutions · Updated 1 September 2026