How a Leading USA Sealing Systems Manufacturer Automated Dynamic Seal Inspection with JIDOKA

Automated multi-stage vision inspection for dynamic seals across 6 variants (60–125 mm), inspecting top/bottom surfaces, IDs, ODs and springs for defects. Runs at 60 parts/min with AI detection and PLC-controlled rejection.

Use Case

Dynamic Seal Defect Detection

Result 1

99% defect detection sensitivity across 6 part variants

Deployments

Kompass AI +Tigris Infrastructure with auto feeder

Result 2

100% Automated Inspection – 1 Part Every Second

Overview

The customer is a leading sealing systems manufacturer in the US, producing precision sealing components for demanding industrial and automotive applications. Consistent seal quality is critical because visual defects in the rubber, trim or spring can affect component performance and downstream reliability.

The manufacturer required an automated inspection solution capable of handling multiple seal variants while inspecting the parts from multiple surfaces. The previous inspection challenge involved identifying a wide range of visual defects consistently, including small flash, tears, missing material, spring-related defects and irregular trimming. Automation was required to improve inspection consistency while keeping pace with production.

The Opportunity

Automating dynamic seal inspection required the system to address several inspection and handling challenges:

Inspecting six different seal variantsacross a diameter range of 60–125 mm.
Inspecting the top surface, bottom face, innerdiameter and outer diameter of each seal.
Capturing the spring region and other criticalsurfaces from the required viewing angles.
Mechanically flipping each part to enableinspection of the opposite side.
Maintaining consistent inspection performanceacross different part sizes.
Rejecting defective parts without interrupting theinspection flow.
Preventing excess grease on the conveyor fromaffecting image acquisition.

‍

JIDOKA's Approach

Multi-Camera Image Acquisition

Station 1 – 5 Cameras

As seals enter the first inspection station, fivecameras capture the bottom face and outer diameter (OD) from multiple views.The cameras are positioned at defined distances and angles to maintainconsistent image quality as parts move through the inspection station.

‍Station 2 – 9 Cameras

After flipping, nine cameras capture the top face,inner diameter (ID) and side profile of the seal. The side-view cameras providededicated coverage for detecting FOG-related defects, while the top and IDviews complete the inspection of the remaining critical surfaces

This two-stage, 14-camera inspection architecture provides comprehensive visual coverage of the seal across both orientations.

‍Automated Part Flipping

A dual pick-and-place mechanism transfers and flips each seal onto a second conveyor. Cameras on the second conveyor capture the bottom face and inner diameter circumference, allowing the system to inspect the remaining critical surfaces without requiring manual handling. This two-stage architecture provides comprehensive visual coverage of the seal.

AI-Based Defect Detection

Captured images are processed on an on-premise industrial computer using a deep-learning-based edge detection algorithm. The AI inspection identifies defects across the defined inspection zones, including:

Loose or flaky flash
Excess flash above 0.7 mm
Splits and tears visible without stretching
Missing material
Irregular or missing FOG
Missing or double springs
Damaged springs
Rusted springs
Tucked lips
Missing grease
Excess wax
Damaged or irregular trim
Short shots
Damaged rubber

Each inspected seal is classified according to the inspection decision.

Real-Time PLC Integration

The AI inspection result is communicated to the machine PLC in real time. When a defective seal is identified, the PLC triggers a pneumatic pusher to remove the NG part from the production flow. This creates an automated sequence from image acquisition to final rejection.

Automatic Part Changeover

The inspection system supports six seal variants with different camera-position requirements. When changing between size ranges, actuators automatically adjust the camera positions to establish the correct working distance for the selected variant. These positions are pre-programmed for the supported part variants. The operator selects the required part variant in the software before starting the production batch.

Production Monitoring and Reporting

A monitoring display provides live production information, including OK and NOT OK part counts by type. Inspection reports can also be downloaded in Excel or CSV format, giving production and quality teams access to inspection data for further analysis.

Conveyor Cleaning

Because seals can carry excess grease, a wiper mechanism beneath each inspection conveyor continuously cleans the conveyor belt. This helps maintain a cleaner inspection surface and supports consistent image acquisition during production. System handling architecture: The feeder-stacker loads seals onto the inspection conveyor, while grippers, a holder unit and pusher mechanism support controlled part handling through the inspection process.

60 Parts per Minute

The inspection machine is designed to handle up to 60 parts per minute,subject to the feeder loading rate.

‍

Six Seal Variants, 60–125 mm

One inspection platform supports six dynamic seal variants across a diameter range of 60–125 mm, with automatic camera adjustment for different part sizes.

‍

Automated 360° Inspection

Top,bottom, inner diameter, outer diameter and spring regions are inspected througha two-stage camera architecture and automated part flipping.

‍

Automated OK/NG Rejection

AI inspection results are communicated to the PLC, which triggers a pneumatic pusher to automatically remove defective parts.

‍

With JIDOKA's automated vision inspection system, the leading US sealing systems manufacturer transformed dynamic seal inspection into a structured, automated process covering multiple surfaces and six part variants. The system combines multi-stage image acquisition, mechanical flipping, AI-based defect detection, automatic camera adjustment and PLC-controlled rejection, with a designed throughput of 60 parts per minute and target inspection performance of 99%sensitivity. The result is a scalable inspection platform designed to improve consistency, automate defect detection and support reliable quality control across production.

‍

Back to All Case Studies