Get a demo

Small Component Assembly Inspection: Catching Missing Snap Rings

6 min read
Worker photographing a snap ring inside an assembled part with a smart glove

Small components such as snap rings, horseshoe washers and plugs are among the hardest assembly details to inspect, because they sit inside the part where a fixed camera cannot see them reliably. The practical answer is to take the camera to the inspection point: the operator photographs each point with a camera on the hand, an AI model returns OK or NOK within seconds, and the result is written to SAP with an evidence photo. TIM INSPECT Component Assembly Inspection works this way with the VEGA-X smart glove.

Why missing snap rings and washers slip through

A snap ring that is not seated in its groove, a washer left out, a plug that was never pushed in: each is a small detail, yet any of them can turn a finished part into a field failure. The defect is rarely visible from the outside. The component sits deep in the assembly, often in a bore or behind a flange, and it looks almost the same whether it is correctly seated or not.

When such a part reaches the customer, the cost is far larger than the component itself: a complaint, a containment action, sorting at the customer's site and a search through records to find which parts are affected. Without a photo and a time stamp for each part, that search is mostly guesswork.

Traditional assembly inspection and its limits

  • Visual checks by the operator. The operator looks into the part and confirms the component is there. It depends entirely on attention, and small rings are easy to miss after hundreds of repetitions.
  • Fixed camera stations. A fixed camera works well when the part is always in the same position against the same background. Here the inspection points are hidden inside the part and the background changes with every shot, so the camera cannot see them consistently.
  • Paper or tick-box records. A signature says that someone checked; it does not show what they saw. When a complaint arrives, there is no image to look at.

How vision AI inspects small components on the line

The solution splits the work between a device that captures and a platform that decides. VEGA-X, the smart glove, scans the barcode, takes the photo and tells the operator the result. The OK or NOK decision is made by TIM INSPECT, using Thread In Motion's own models trained on industrial images.

  1. Scan the barcode. The operator scans the part with VEGA-X. The part is identified by its part code and traceability code.
  2. Capture the inspection points. The operator holds a hand to the snap ring, washer and plug points in turn, and the glove photographs each one up close and at the right angle.
  3. Get the result instantly. TIM INSPECT returns OK or NOK. The result comes back to the glove as vibration, sound and light, so both hands stay on the part and nobody has to look at a screen.
  4. Write the record. The result goes to the management panel and SAP with its evidence photo. A part with a NOK does not move on.

Take a snap ring as an example. The model looks for the hole and both edges of the ring. If the hole is found but only one edge is detected, the ring is not seated and the decision is NOK. The glove vibrates and its light turns red, the operator seats the ring, photographs the same point again, and only an OK lets the part continue.

Glare, oil and shadows

Images on an assembly line are rarely clean. They are very close-up greyscale shots taken with glare, oil and shadows in the frame. The models are trained on industrial images of exactly this kind. For a new part, the scope is assessed using sample images from your own floor before anything is promised.

Results from a deployment

At an automotive supplier, component assembly inspection has completed more than 12.8 million inspections with zero defective parts reaching the customer, at under 4 seconds per inspection end to end, from photo to feedback. These figures come from the Thread In Motion project report for that deployment.

What to plan before you start

  • Inspection points. List the components to check on each part and the order in which the operator should present them. Start with the points that have caused complaints.
  • Part identification. The barcode or traceability code is what ties each photo to a part. Make sure it is readable at the station.
  • Target system. In the deployment above, results are written to SAP per part. Depending on your setup, TIM INSPECT can connect to SAP, MES, ERP, WMS or PLC systems; agree on the scope before installation.
  • NOK handling. Decide what happens to a NOK part: rework at the station, a hold area, or a second check. The record shows how many times a part was reworked.
  • Using the records. Each record holds part code, traceability code, result, date and time, and the evidence photo. Records can be searched, filtered by date and exported to Excel, which is what you need when a complaint arrives.

Quick-connect fittings raise a similar question of whether a connection is fully seated; see Quick-Connect Verification. Both run on the same TIM INSPECT platform for operational vision AI.

FAQ

Why can't a fixed camera do this inspection?

The inspection points sit inside the part and the background changes with every shot, so a fixed camera cannot see them reliably. A camera on the hand reaches the point up close and at the right angle.

Does the glove make the decision?

No. VEGA-X scans, captures and reports the result. The decision is made by TIM INSPECT with Thread In Motion's own models.

Does the inspection slow the line down?

The commitment for VEGA-X is under 4 seconds end to end, from photo to feedback, and the operator reads the result from the vibration with both hands on the part.

Which components can be inspected?

The deployment above covers snap rings, horseshoe washers and plugs. Other components are assessed with sample images from your floor.

Which systems receive the results?

Results can be written to SAP, MES, ERP, WMS or PLC systems, depending on what you use.

Next step

If you have a component that keeps causing complaints, request a demo and we can look at your inspection points and sample images together.

Share this article LinkedIn X
Written by Zeynep Fert
Thread in Motion

See TIM on your own floor

Wearable scanners, Physical AI wearables and vision AI for frontline operations. Let us show you how they fit your process.

Get a demo

Keep reading

Chat on WhatsApp WhatsApp'tan yazın