Intelgic · Talk to Us

Rivet and Spot-Weld Inspection with Visual AI: Presence, Position and Quality

Back to Knowledge base
Intelgic · Technical Guide Rivet & Spot-Weld Inspection Automotive & Aerospace Manufacturing

Rivet and Spot-Weld Inspection with Visual AI: Presence, Position and Quality

Intelgic automates rivet and spot-weld inspection by combining high-resolution industrial cameras, specialized lighting, robots or collaborative robots, state-of-the-art AI, and the Certainty inspection platform.

Intelgic · Manufacturing Automation Published: 2026/09/17 Rivets · Spot Welds · Robotics · AI Vision
00 · Introduction

Rivet and Spot-Weld Inspection with Visual AI: Presence, Position and Quality

Intelgic automates rivet and spot-weld inspection by combining high-resolution industrial cameras, specialized lighting, robots or collaborative robots, state-of-the-art AI, and the Certainty inspection platform.

The system captures an image of every location where a rivet or spot weld is expected. Intelgic's AI then checks three fundamental quality conditions: presence, position, and visible quality.

Presence — Is the required rivet or spot weld present?
Position — Is it located within the permitted area?
Quality — Does its visible appearance meet the defined acceptance criteria?

For large or geometrically complex components, a robot moves the camera and lighting assembly from one inspection region to another. For smaller or flatter parts, fixed cameras, indexing systems, line-scan cameras, or robotic part handling may be used.

Certainty automatically loads the correct inspection recipe for each component variant, coordinates image acquisition and AI analysis, creates a digital defect map, and sends inspection results to the existing PLC, MES, ERP, or quality-management system.

01 · Guide Section

Challenges

Although rivets and spot welds are visible manufacturing features, their automated inspection presents several imaging and quality-control challenges.

Rivet and Spot-Weld Inspection with Visual AI

Large numbers of inspection locations

A component may contain hundreds or thousands of rivets and spot welds. Each location must be checked against the correct design or inspection map. A feature can be missing even when all surrounding locations are correct, so the system must verify every required position rather than relying only on an overall appearance check.

Similar-looking features

Rivets, spot welds, holes, surface marks, sealant, and reflections may have similar shapes in an image. The AI must distinguish an actual fastening or joining feature from an empty hole, tooling mark, stain, weld spatter, or other visual pattern.

Reflective metal surfaces

Bare aluminium, coated steel, stainless steel, painted panels, and metallic fasteners can create glare. Reflections may hide a small crack, distort the apparent edge of a rivet, or make a spot weld appear larger or smaller.

Variations in acceptable appearance

Acceptable rivets and spot welds do not always look identical. Their appearance can vary because of material, coating, surface texture, electrode condition, tooling, sealant, lighting angle, part geometry, and normal process variation. The inspection system must recognize acceptable variation without overlooking genuine defects.

Complex component geometry

Rivets and spot welds may be distributed across flat, curved, angled, vertical, or recessed surfaces. One fixed camera cannot always image every location at the required angle and resolution.

Multiple part variants

Different product models may have different rivet patterns, spot-weld maps, feature counts, surface geometries, materials, acceptance limits, and inspection regions. The correct inspection configuration must be applied automatically to each variant.

02 · Guide Section

How Intelgic Addresses the Challenges

Intelgic designs the imaging and automation architecture around the component geometry, joining process, defect requirements, production rate, and available inspection time.

The component is divided into multiple inspection regions. For each region, Certainty defines the camera position, lighting condition, expected feature locations, AI model, and inspection limits.

Controlled imaging

The inspection station uses controlled illumination to reduce the influence of changing ambient factory light. Depending on the surface and defect type, Intelgic may use:

Low-angle dark-field lighting
Diffuse dome lighting
Coaxial illumination
Directional bar lights
Polarized lighting
Multi-directional illumination
Structured or patterned lighting

The inspection may take place inside a dark or optically enclosed cell when the surface is highly reflective or the required defects are particularly small.

Multi-light image capture

One lighting direction may reveal a scratch or raised edge while another provides a better view of weld shape or surface indentation. Certainty can capture several images of the same region under different lighting conditions. Intelgic's AI analyzes the complementary images to separate actual physical features from glare, shadows, and reflections.

Robotic camera positioning

A robot or cobot moves the camera and lighting assembly to each inspection region. The robot maintains the required:

Camera-to-surface distance
Viewing angle
Focus
Field of view
Lighting geometry
Image overlap
Sensor orientation

For large components, the robot may be installed on a linear track. Multiple robots can also be used where one robot cannot provide the necessary coverage or cycle time.

Part localization

Reference cameras, laser sensors, or 3D sensors determine the component's actual position. Certainty can align the stored inspection map with the physical part and compensate for permitted fixture or loading variation. This prevents normal part-position changes from being mistaken for incorrectly positioned rivets or welds.

03 · Guide Section

Inspection Configurations

Robot-mounted camera

A robot carries the camera and lighting system around a stationary component. This configuration is suitable for:

Large panels
Aircraft structures
Automotive body assemblies
Doors
Frames
Curved sheet-metal components
Welded or riveted structures with multiple surfaces

Robot moves the component

For smaller parts, a robot can pick up the component and present its different surfaces to stationary cameras and lights. This can provide repeatable imaging while keeping the optical equipment fixed.

Fixed multi-camera station

Several cameras capture different component surfaces simultaneously or in a short sequence. This approach can support high-speed inspection when part position and geometry are consistent.

04 · Guide Section

How the Inspection Process Works

01Part identification — The system identifies the component using a barcode, data-matrix code, RFID tag, PLC signal, MES work order, or verified operator selection.
02Recipe loading — Certainty loads the inspection recipe for the correct component model and variant. The recipe can define expected rivet locations, expected spot-weld locations, feature type at each position, robot path, camera positions, focus and exposure, lighting sequence, inspection regions, AI models, position tolerances, visible quality criteria, and reporting requirements.
03Part localization — The system identifies reference features and calculates the component's actual position. Certainty aligns the digital feature map with the physical component before inspection begins.
04Robotic positioning — The robot moves the camera and lighting assembly to the first region. For a fixed-camera system, a conveyor or indexing mechanism positions the component instead.
05Image acquisition — Certainty activates the required lighting and captures one or more images. Images may be taken under different lighting directions when necessary to expose surface defects and reduce reflections.
06Image-quality verification — The software verifies that every image is in focus, correctly exposed, properly aligned, free from unacceptable glare, captured at the expected location, and suitable for AI analysis. An invalid image can be captured again or marked for review rather than automatically passed.
07Presence inspection — The AI compares the image with the expected rivet and spot-weld map, identifying present features, missing features, unexpected or additional features, obstructed inspection locations, and locations requiring review.
08Position inspection — The system calculates the center or reference point of each detected rivet or spot weld, then compares the actual location with the nominal position and permitted tolerance zone.
09Visible quality inspection — Intelgic's AI analyzes the appearance of each feature for trained defect conditions. Rule-based measurements and 3D data may be combined with AI when the inspection requires quantitative dimensions.
10Defect mapping and reporting — Every result is linked to its physical location on the component. Certainty generates the overall inspection status and sends the required information to the PLC, MES, ERP, quality system, or cloud analytics platform.
05 · Guide Section

Presence Inspection

Presence inspection verifies that every required joining feature exists. The system compares detected rivets and welds with a digital reference map derived from engineering data, CAD information, inspection plans, approved reference components, or manually configured feature coordinates.

Possible presence results include:

Correct feature present
Rivet missing
Spot weld missing
Empty hole
Extra rivet
Extra weld mark
Wrong feature type
Feature obstructed
Image invalid

The system can also count the total number of rivets and spot welds and compare the result with the expected count.

Counting alone is not sufficient. A component could contain the correct total number but still have one missing feature and one unexpected feature elsewhere. Location-by-location verification provides a more reliable result.

06 · Guide Section

Position Inspection

Position inspection verifies whether the detected feature falls within the specified tolerance zone. The software can measure:

Position Inspection
Feature center
Offset from nominal location
Distance from an edge
Distance between adjacent features
Row alignment
Pattern spacing
Position relative to a hole, flange, seam, or reference feature

For spot welds, an incorrectly located weld may fall too close to an edge or outside the intended sheet overlap. For rivets, the system may detect an off-center or misaligned installation relative to the expected hole or fastener pattern.

Measurement accuracy depends on camera calibration, image resolution, lens distortion, part localization, viewing angle, and surface geometry.

07 · Guide Section

Quality Inspection

Quality inspection evaluates the visible condition of the rivet or spot weld. Because rivets and spot welds have different defect mechanisms, Certainty can apply a separate AI model and inspection logic to each feature type.

The system can also use location-specific criteria. A critical region may have different inspection requirements from a noncritical or cosmetic area.

08 · Guide Section

State-of-the-Art AI for Defect Detection

Traditional machine vision works well for features that can be evaluated using fixed limits for size, position, circularity, contrast, or color. Rivet and spot-weld defects are often irregular, and their appearance can also change with material, coating, lighting, sealant, and manufacturing variation.

Intelgic's AI learns these visual patterns from representative images of acceptable and defective components. The AI may perform:

Feature detection

Finds each rivet or spot weld and associates it with the correct expected location.

Classification

Determines whether the feature is acceptable or belongs to a trained defect category.

Segmentation

Identifies the pixels associated with the feature or defect, allowing its visible dimensions and area to be calculated.

Anomaly detection

Flags a location that differs from validated examples of acceptable features, including unusual conditions not included in a predefined defect class.

Multi-image analysis

Compares images captured under different lighting conditions to distinguish defects from reflections and surface variation.

Confidence evaluation

Assigns a confidence score that can be used to route uncertain results for qualified review.

09 · Guide Section

Combining AI with Rule-Based Vision and 3D Measurement

The most effective inspection architecture may combine several approaches. For example, for a rivet:

AI detects and classifies a damaged rivet.
Rule-based vision calculates its position and head diameter.
A 3D sensor measures its height or flushness.
Certainty compares the results with the configured acceptance limits.

For a spot weld:

AI detects an abnormal visible weld condition.
Rule-based software measures the visible electrode impression.
A 3D sensor measures indentation depth.
Certainty records the result against the weld's specified location.

This hybrid approach uses AI for complex defect appearance and conventional metrology for calibrated measurements.

10 · Guide Section

Certainty Inspection Platform

Certainty manages the complete rivet and spot-weld inspection workflow. It coordinates:

Part identification
Recipe selection
Digital feature maps
Robot movement
Camera triggering
Lighting sequences
Image acquisition
Image preprocessing
AI inference
Feature counting
Position measurement
Defect classification
Pass/fail/review logic
Defect mapping
Report generation
System integration
Data storage
Quality analytics
11 · Guide Section

Recipe Management for Different Variants

Certainty can maintain an individual inspection recipe for every component variant. The correct recipe can be loaded automatically using the part identity or manufacturing order. A recipe may control:

Part geometry
Rivet and spot-weld map
Expected feature count
Robot or cobot path
Camera angle
Working distance
Focus
Exposure
Lighting sequence
Inspection regions
AI model
Position tolerances
Quality thresholds
Reporting format

Automatic recipe selection allows one inspection cell to process components with different dimensions, patterns, materials, and acceptance requirements.

12 · Guide Section

Integration with Existing Systems

Certainty can be integrated with existing factory automation and manufacturing software.

PLC integration

The platform can exchange:

Part-present signals
Part identity
Fixture status
Inspection start
Recipe confirmation
Robot status
Inspection completion
Pass/fail/review results
Rework-routing commands
Fault and alarm information

MES integration

MES connectivity can support:

Work-order retrieval
Automatic recipe selection
Serial-number tracking
Part genealogy
Inspection-result storage
Rework workflows
Production reporting
Quality traceability
Recipe revision control

Additional connectivity

Certainty can also connect with:

ERP systems
SCADA
Quality-management systems
Factory databases
Data historians
Cloud platforms
Customer-specific applications
13 · Guide Section

Cloud-Based Quality Inspection Analytics

Inspection data can be transferred to a secure cloud analytics environment, subject to the manufacturer's cybersecurity and data-governance policies.

Quality dashboards

Quality dashboards may display:

Components inspected
Pass and fail rates
Missing rivets
Missing spot welds
Position deviations
Defects by category
Defects by component variant
Defects by production line
Defects by shift
Inspection cycle time
Rework frequency
Quality trends over time

Defect heat maps

Aggregated results can show where rivet or spot-weld defects occur most frequently on the component. Repeated defects in one area may indicate issues involving:

Riveting equipment
Welding electrodes
Robot programming
Fixtures
Sheet alignment
Part fit-up
Material handling
Process settings
Tool wear

Image-level traceability

Authorized users can review original images, annotated defects, measured positions, AI confidence, inspection recipes, and production history.

Multi-site quality monitoring

Cloud analytics can provide authorized teams with consolidated quality information across production lines, plants, and suppliers. An on-premises or hybrid architecture can be used where image or production-data storage in the cloud is restricted.

14 · Guide Section

Digital Inspection Records

For each component, Certainty can record:

Part number
Serial number
Component variant
Work order
Inspection date and time
Recipe version
AI-model version
Expected feature count
Actual feature count
Individual rivet results
Individual spot-weld results
Position deviations
Defect category
Defect dimensions
Defect location
AI confidence
Original images
Annotated images
Pass/fail/review status
Reinspection result
Rework history
15 · Guide Section

Rivet Defects That Can Be Detected

Depending on the validated imaging system, Intelgic's AI can inspect for visible conditions such as:

Missing rivets

The expected location contains no rivet or contains an empty hole.

Damaged rivet heads

Visible damage may include:

Cracks
Chips
Gouges
Dents
Tool marks
Deformation
Broken or incomplete heads

Incorrect rivet type

The visible head shape, dimensions, color, marking, or other feature differs from the expected rivet.

Raised or recessed rivets

A rivet may sit above or below the surrounding surface. Appearance-based inspection can identify an abnormal condition, while calibrated 3D sensing may be required for precise flushness measurement.

Tilted or misaligned rivets

The head may appear asymmetrical or incorrectly aligned with the panel surface.

Cracks around rivets

The system may detect visible cracks extending from the fastener location when the crack width, surface finish, lighting, and image resolution provide sufficient contrast.

Sealant anomalies

Visible conditions may include missing, excessive, discontinuous, or contaminated sealant around a rivet.

16 · Guide Section

Spot-Weld Defects That Can Be Detected

Depending on the application, surface imaging can inspect for:

Missing spot welds

The expected location contains no visible weld impression.

Incorrectly positioned welds

The weld is offset from its specified coordinate or falls outside the permitted joining area.

Expulsion and spatter

Excessive material ejection may produce visible spatter, pits, or irregular surface conditions.

Burn-through

Excessive heat may create a hole or severe local surface damage.

Excessive indentation

The electrode impression appears deeper or more severe than the approved visible condition. Quantitative depth generally requires 3D measurement.

Irregular weld impression

The weld mark may have an abnormal shape, size, edge, or surface texture.

Visible cracking

Surface-breaking cracks may be detected when the imaging conditions provide adequate contrast.

Electrode-related abnormalities

Changes in the weld impression may indicate electrode wear, contamination, misalignment, or other process variation.

Surface imaging does not directly determine internal weld fusion, nugget size, internal rivet formation, or joint strength. Where these properties must be verified, another validated method may be required alongside visual inspection.

Looking to automate rivet and spot-weld inspection? Contact Intelgic to discuss a robotic Visual AI system for presence, position, and visible quality inspection powered by the Certainty platform.

17 · Guide Section

Frequently Asked Questions

Can one system inspect both rivets and spot welds? +

Yes. Certainty can apply separate inspection logic and AI models to rivets and spot welds while managing both within the same component recipe and inspection report.

Can AI detect a missing rivet or spot weld? +

Yes. The AI compares every expected location with the captured image and identifies missing, additional, or obstructed features.

Can the system measure feature position? +

Yes. A calibrated vision system can calculate the feature center and compare it with the nominal coordinate and permitted tolerance zone.

Can AI inspect rivet and weld quality? +

AI can inspect defined visible conditions such as damaged rivet heads, abnormal seating, weld expulsion, burn-through, irregular impressions, and visible cracks.

Can visual inspection verify the internal strength of a joint? +

No. Surface imaging does not directly determine internal weld fusion, nugget size, internal rivet formation, or joint strength. Other validated methods may be required.

Can one inspection cell handle multiple part variants? +

Yes. Certainty can store variant-specific rivet and weld maps, robot paths, camera settings, lighting sequences, AI models, and inspection limits.

Can defects be displayed on a component map? +

Yes. Certainty can link each result to its physical location and display missing, misplaced, or damaged features on a digital defect map.

Can Certainty integrate with an existing PLC and MES? +

Yes. Certainty can exchange part identity, recipe, machine status, inspection results, alarms, traceability, and rework information with existing systems.

Are inspection results available through cloud dashboards? +

Yes. Subject to the manufacturer's data policies, dashboards can display defect trends, heat maps, pass/fail rates, images, feature-level results, and production-quality metrics.

Head Quarter

17352 Murphy Ave Suite 101
Irvine, CA 92614
Phone: (949) 317-2420

Branch Office

5ES7-H, 5Th Floor, East Tower, Mani Casadona, Plot no -IIF/04, Action Area -IIF, Newtown, Kolkata-700160

Certificate

intelgic ISO 27001 certification

Social links

©2026 Intelgic Inc. All Rights Reserved.