Intelgic · Talk to Us

Robotic Defect Detection on Car Doors: Dents, Scratches, Paint Defects and Panel Deformation

Back to Knowledge base
Intelgic · Technical Article Automotive Inspection AI Machine Vision

Robotic Defect Detection on Car Doors:
Dents, Scratches, Paint Defects
and Panel Deformation

Intelgic combines robotic motion, controlled industrial lighting, high-resolution 2D imaging, 3D measurement and AI-powered analysis to inspect car doors from multiple viewpoints.

Intelgic · Irvine, CA Published 2026/09/02 Automotive · Robotics · 2D/3D Vision · Defect Detection
01 · Introduction

Robotic Defect Detection on Car Doors: Dents, Scratches, Paint Defects and Panel Deformation

A car door is both a structural assembly and one of the vehicle’s most visible exterior surfaces. A small dent, scratch, paint inclusion or contour error can affect customer perception, create fit-and-finish problems or lead to expensive rework after final assembly.

Inspecting a car door is difficult because its surface is large, curved and highly reflective. Defects that disappear under ordinary factory lighting may become obvious when the vehicle reaches daylight or showroom illumination. At the same time, the inner structure contains folds, openings, welds, brackets and assembly features that cannot be inspected from one camera position.

Intelgic combines robotic motion, controlled industrial lighting, high-resolution 2D imaging, 3D measurement and AI-powered analysis to inspect car doors from multiple viewpoints. The system can detect cosmetic defects, measure panel deformation and create a traceable quality record for every inspected door.

02 · Why Car Door Inspection Is Challenging

Why Car Door Inspection Is Challenging

A door changes significantly as it moves through manufacturing. It may be inspected as:

  • An outer stamped panel
  • An inner stamped structure
  • A hemmed door assembly
  • A welded assembly
  • A primed component
  • A painted Class-A panel
  • A fully assembled door
  • A door fitted to the vehicle body

Each stage introduces different defects and requires a different imaging strategy. The main inspection challenges include:

  • Large surface area
  • Complex three-dimensional geometry
  • Highly reflective painted surfaces
  • Different paint colors and finishes
  • Subtle, low-contrast defects
  • Hidden and recessed inspection regions
  • Several door models on one line
  • Tight automotive cycle times
  • Environmental vibration and ambient light
  • The need to distinguish cosmetic defects from geometric deformation

A single camera and standard ring light are rarely sufficient. The inspection system must control how light interacts with the door surface and move the sensor to maintain a suitable angle across its changing contours.

03 · Defects Intelgic Can Detect on Car Doors

Defects Intelgic Can Detect on Car Doors

The inspection scope depends on the manufacturing stage, required minimum defect size and customer acceptance criteria.

Dents and dings

Dents create local changes in surface shape. Some are visually obvious, while shallow dents may become visible only when a reflected light pattern crosses the affected area. The system can be configured to detect:

  • Handling dents
  • Pressing or stamping dents
  • Local depressions
  • Raised bumps
  • Edge dings
  • Hail-type damage
  • Low spots
  • High spots
  • Dents around handles or feature lines

Shallow deformation is generally inspected using structured reflection, deflectometry or high-resolution 3D measurement rather than relying on a normal photograph alone.

Scratches and scuffs

Scratches can range from light clear-coat marks to deep damage exposing the underlying paint or metal. Intelgic’s system can detect defined classes of:

  • Fine scratches
  • Deep scratches
  • Scuff marks
  • Abrasion
  • Drag marks
  • Polishing marks
  • Edge scratches
  • Handling damage

Low-angle and dark-field illumination can make scratches appear bright against the surrounding surface. Multiple lighting directions help reveal defects regardless of their orientation.

Paint and coating defects

Paint defects can be geometric, color-related or texture-related. Typical examples include:

  • Dirt inclusions and nibs
  • Dust contamination
  • Fibres
  • Paint runs and sags
  • Orange peel
  • Fisheyes
  • Craters
  • Pinholes
  • Blisters
  • Bubbles
  • Uneven gloss
  • Mottling
  • Cloudiness
  • Color inconsistency
  • Insufficient or excessive visible coverage
  • Overspray
  • Unpainted regions
  • Clear-coat defects
  • Burn-through after polishing
  • Water spots and stains

Automotive vision systems commonly inspect paint surfaces for defects such as scratches, dents, runs, dirt and orange-peel texture. Automotive machine-vision applications Visible inspection does not directly measure paint-film thickness unless the cell includes a suitable dedicated thickness sensor. Surface appearance and coating thickness should therefore be treated as separate inspection requirements.

Panel deformation and waviness

A door can look acceptable in a local image while its overall geometry falls outside specification. 3D measurement can detect:

  • Global panel bow
  • Local warpage
  • Surface waviness
  • Twisting
  • Incorrect curvature
  • Feature-line deformation
  • Hemming distortion
  • Edge displacement
  • Flange deformation
  • Incorrect contour
  • Distortion around openings
  • Assembly-induced deformation

The measured surface can be compared with a CAD model, approved master panel or defined tolerance zones.

Edge, hole and feature defects

The system can also inspect:

  • Damaged or rolled edges
  • Incorrect trim profiles
  • Hole presence and position
  • Slot geometry
  • Handle openings
  • Mirror-mounting features
  • Flange position
  • Hem quality
  • Cracks near formed regions
  • Missing or distorted cutouts

Assembly defects

For inner or completed door assemblies, additional cameras can verify:

  • Missing clips
  • Missing fasteners
  • Incorrect brackets
  • Wiring-harness routing
  • Grommet presence
  • Seal placement
  • Adhesive or sealant presence
  • Speaker and actuator installation
  • Handle components
  • Loose foreign objects
  • Incorrect component orientation

Intelgic also supports inspection of foreign objects and assembly anomalies inside car-door cavities. Intelgic car-door inspection automation

04 · Why Painted Car Doors Need Specialized Imaging

Why Painted Car Doors Need Specialized Imaging

A glossy door behaves like a curved mirror. The camera often sees reflections of lights, equipment and the inspection enclosure rather than the paint surface itself. A normal image may therefore contain:

  • Bright glare
  • Dark reflected objects
  • Moving highlights
  • Reflections of the robot
  • Intensity changes caused by surface curvature
  • Different responses from different paint colors

AI cannot compensate for a defect that was never made visible. The optical system must first create an image in which the defect has reliable contrast. Intelgic selects the camera, lens and illumination geometry according to the target defect rather than using one lighting setup for every inspection.

05 · Imaging Methods for Car Door Inspection

Imaging Methods for Car Door Inspection

Diffuse illumination

Large diffuse lights create a more uniform image of glossy paint and reduce sharp reflections. They are useful for general surface appearance, color and contamination inspection.

Dark-field and low-angle lighting

Light directed at a shallow angle highlights scratches, particles, raised paint defects and sharp surface discontinuities. Because a scratch reflects light differently depending on its direction, the system may capture several images with lights activated from different sides.

Structured reflection

A known stripe, grid or sinusoidal pattern is reflected from the painted surface. A smooth panel produces an orderly reflected pattern, while dents, waviness and coating irregularities distort it. This approach is particularly useful for glossy Class-A surfaces because the reflection itself becomes the measurement signal. Robot-guided automotive paint-inspection systems use controlled sensor positioning and multiple illumination methods to inspect curved painted surfaces. Robot-guided paint inspection example

Polarized imaging

Polarizing filters can reduce selected reflections and improve the visibility of some paint, stain and surface defects. Their effectiveness depends on the coating and lighting angle.

Backlighting

Backlighting produces a high-contrast silhouette and is useful for evaluating:

  • Outer contour
  • Holes and slots
  • Edge shape
  • Trim conditions
  • Certain gaps

3D laser profiling

A laser profiler projects a line onto the door. The sensor observes its displacement and calculates a cross-sectional height profile. As the robot moves the profiler across the panel, consecutive profiles are combined into a three-dimensional surface map.

Multi-image acquisition

The same region may be captured several times using different lighting directions, exposures or sensor positions. Combining these images helps separate actual defects from harmless reflections.

06 · How Intelgic’s Robotic Car Door Inspection Works

How Intelgic’s Robotic Car Door Inspection Works

1. Door arrival and identification

The door arrives through a conveyor, carrier, skid or manual loading station. Its model, side, color and production identity can be received from:

  • PLC
  • MES
  • Barcode
  • QR code
  • RFID
  • Production schedule

The system automatically loads the appropriate inspection recipe and robot path.

2. Position and orientation verification

The software locates the door or verifies its fixture position. Reference features may be used to compensate for permitted loading variation. For moving carriers, the system can synchronize inspection with conveyor position or use tracking data from the line controls.

3. Robot path execution

One or more robots move cameras, lights or 3D sensors along programmed paths around the door. The robot maintains the required:

  • Camera-to-surface distance
  • Viewing angle
  • Sensor orientation
  • Lighting geometry
  • Scanning speed
  • Overlap between inspection regions

For complex curved panels, the path can be developed from CAD data and refined during commissioning.

4. Multi-angle image capture

At each inspection pose, the system captures one or more images. Lighting channels may be triggered sequentially to reveal different defect types. Fixed cameras can be added for inspection tasks that benefit from simultaneous acquisition, such as overall part verification or edge measurement.

5. AI defect analysis

Intelgic’s Live Vision AI software analyzes the captured images and identifies suspicious regions. The software can:

  • Detect defects
  • Classify defect type
  • Segment the defect area
  • Measure visible defect dimensions
  • Assign a confidence score
  • Compare the result with location-specific tolerances

Intelgic’s platform supports detection of scratches, dents, contamination, discoloration and other visual anomalies. Intelgic Live Vision AI

6. Three-dimensional analysis

The 3D system analyzes panel geometry independently of color and many visible-light variations. It can calculate:

  • Dent depth
  • Raised-defect height
  • Panel flatness
  • Local curvature
  • Waviness
  • Feature-line deviation
  • Edge position
  • Gap
  • Flushness
  • Deviation from CAD or master data

7. Sensor-data fusion

A 2D defect and a 3D deviation can be registered to the same door coordinate system. This helps distinguish defects such as:

  • A dark mark with no height change, which may be a stain
  • A local height change with little color difference, which may be a dent
  • A bright feature with positive height, which may be a paint nib
  • A long directional surface response, which may be a scratch
  • Broad geometric deviation, which may indicate panel deformation

8. Pass, fail or review decision

The system evaluates each defect according to its:

  • Type
  • Size
  • Depth or height
  • Location
  • Severity
  • Confidence
  • Customer-specific acceptance rule

A small defect may be accepted in a hidden region but rejected on a customer-facing Class-A surface.

9. Traceability and production action

The system can send results to a PLC, MES or quality database. Defective doors may be:

  • Diverted to rework
  • Marked automatically
  • Assigned to a manual review station
  • Prevented from advancing to the next process
  • Linked to a repair instruction
07 · 2D Vision vs. 3D Inspection for Car Doors

2D Vision vs. 3D Inspection for Car Doors

Inspection requirement2D vision3D measurement
Color variationExcellentLimited
Stains and contaminationExcellentOnly if geometry changes
Fine scratchesExcellent with suitable lightingMay miss very shallow scratches
DentsIndirectly visible through reflected patternsDirect depth measurement
Paint nibsGoodMeasures raised geometry
Orange peelGood with specialized optical methodsPossible with sufficient resolution
Panel warpageLimitedExcellent
Edge and contour measurementGood for silhouettesExcellent for full geometry
Print and label inspectionExcellentNot normally required
Gap and flushnessLimited in 2DExcellent

The technologies are complementary. For car-door inspection, combining appearance imaging with 3D measurement usually provides more complete defect coverage than using either method alone.

08 · Inline and Offline Inspection Configurations

Inline and Offline Inspection Configurations

Inline robotic inspection

The door is inspected within the production flow, either stationary for a short indexed cycle or while moving under controlled conditions. Advantages include:

  • Inspection of every door
  • Immediate process feedback
  • Automatic traceability
  • Faster containment of production problems

The available line cycle determines the number of robot paths, images and scans that can be completed.

Offline inspection cell

The door is diverted or manually loaded into a dedicated inspection cell. Advantages include:

  • Longer inspection time
  • More sensor viewpoints
  • Higher-resolution scanning
  • Easier investigation of failed parts
  • Suitability for audit and rework inspection

Hybrid inspection

A fast inline station can screen every door, while suspicious products are sent to a detailed robotic cell for higher-resolution analysis.

09 · Inspection at Different Manufacturing Stages

Inspection at Different Manufacturing Stages

After stamping

The system can detect:

  • Dents
  • Cracks
  • Splits
  • Wrinkles
  • Edge deformation
  • Hole and slot errors
  • Surface waviness
  • Springback-related geometry changes

After hemming and welding

Inspection can cover:

  • Hem deformation
  • Weld presence
  • Assembly alignment
  • Distortion
  • Flange position
  • Sealant application
  • Inner-to-outer panel relationship

After painting

The system can inspect:

  • Scratches
  • Dirt inclusions
  • Runs and sags
  • Craters
  • Orange peel
  • Pinholes
  • Gloss variation
  • Color inconsistency
  • Dents visible through the finish

After final door assembly

Inspection can verify:

  • Trim and hardware presence
  • Handle installation
  • Wiring
  • Seals
  • Labels
  • Fasteners
  • Foreign objects
  • Final surface condition

After body fitment

The system can measure:

  • Door-to-fender gap
  • Door-to-quarter-panel gap
  • Door-to-sill alignment
  • Flushness
  • Feature-line continuity
  • Door sag
  • Fit relative to surrounding body panels
10 · AI Training and Validation

AI Training and Validation

AI models require representative production data. Training images should cover:

  • Different door models
  • Left- and right-hand doors
  • Paint colors
  • Metallic and non-metallic finishes
  • Acceptable process variation
  • Naturally occurring defects
  • Different defect positions
  • Relevant defect sizes
  • Normal reflections and environmental variation

The system should be validated separately for critical defect classes. Useful performance measures include:

  • Defect-detection rate
  • Missed-defect rate
  • False-reject rate
  • Classification accuracy
  • Defect-location accuracy
  • Measurement repeatability
  • Cycle time

A single generic accuracy percentage does not describe performance across every paint color, defect type and door geometry.

11 · Factors Affecting Detection Performance

Factors Affecting Detection Performance

Detection capability depends on:

  • Minimum defect size
  • Defect depth or height
  • Defect orientation
  • Paint color and gloss
  • Metallic or pearlescent finish
  • Surface curvature
  • Camera resolution
  • Lighting geometry
  • Robot repeatability
  • Conveyor movement
  • Environmental contamination
  • Required cycle time

Representative feasibility testing is therefore an essential part of system design.

12 · Benefits of Automated Robotic Car Door Inspection

Benefits of Automated Robotic Car Door Inspection

Consistent quality decisions

The same inspection criteria are applied to every door, independent of operator, shift or fatigue.

Multi-angle coverage

Robots position sensors around large and curved surfaces that cannot be covered from one fixed viewpoint.

Reduced rework cost

Defects can be detected before additional components are installed or before the door reaches final vehicle assembly.

Faster root-cause analysis

Defect images and locations help identify problems associated with stamping tools, hemming equipment, paint processes or material handling.

Multi-variant flexibility

Robot paths and inspection recipes can support several door models within the same cell.

Digital traceability

Each door can be linked with its inspection images, defect map, measurements and final decision.

13 · Intelgic’s End-to-End Solution

Intelgic’s End-to-End Solution

Intelgic’s robotic car-door inspection system can include:

  • Industrial or collaborative robots
  • High-resolution 2D cameras
  • 3D laser profile sensors
  • Structured-reflection imaging
  • Diffuse and directional lighting
  • Fixed and robot-mounted cameras
  • Controlled inspection enclosures
  • Live Vision AI software
  • CAD and master-part comparison
  • Defect classification and measurement
  • Multi-model recipe management
  • PLC and MES integration
  • Rework-station communication
  • Image storage and analytics

The system architecture is developed around the actual door, production stage, target defects and available cycle time.

14 · From Manual Visual Checks to Digital Surface Quality

From Manual Visual Checks to Digital Surface Quality

Car-door inspection cannot be solved by AI software alone. A successful system must first make dents, scratches and paint defects visible through carefully engineered lighting and sensor positioning. Intelgic combines robotic flexibility with 2D appearance imaging, 3D measurement and AI analysis. This enables manufacturers to inspect complex door surfaces consistently, quantify panel deformation and create an actionable defect record for every product. The result is not simply an automated pass-or-fail station. It is a source of process information that can help stamping, body, paint and final-assembly teams prevent recurring defects.

15 · Frequently Asked Questions

Frequently Asked Questions

What car-door defects can Intelgic detect? +

Depending on the configuration, the system can detect dents, scratches, scuffs, paint inclusions, runs, sags, craters, stains, pinholes, orange peel, edge damage, panel deformation and assembly defects.

Why are painted car doors difficult to inspect? +

Glossy paint reflects the inspection environment. Surface curvature also changes those reflections across the door. Specialized lighting and controlled camera angles are needed to distinguish real defects from normal highlights.

Can a 2D camera measure dent depth? +

A 2D system can reveal a dent through reflected-pattern distortion, but it does not directly measure depth. A calibrated 3D sensor is normally used when dent depth or panel deformation must be quantified.

Can the system detect fine scratches? +

Yes, when the scratch is large enough for the selected resolution and produces sufficient contrast under the configured lighting. Multiple low-angle illumination directions may be used to detect scratches with different orientations.

Can Intelgic detect paint color variation? +

Yes. Color inspection requires calibrated cameras, stable illumination and agreed acceptance limits. Metallic, pearlescent and highly glossy finishes require application-specific validation.

Can the system detect orange peel? +

Surface-texture irregularity can be inspected using specialized lighting or structured-reflection methods. The required optical resolution and acceptance method must be established using representative samples.

Does the system measure paint thickness? +

Not with standard 2D imaging alone. Paint-thickness measurement requires an appropriate dedicated sensor or test method. Intelgic can integrate additional sensors when thickness is part of the inspection requirement.

Can the door be inspected while moving? +

Yes, in suitable line configurations. The system may track a continuously moving carrier or inspect the door during an indexed stop. Achievable coverage depends on conveyor speed, robot motion and image-acquisition time.

How many robots are required? +

The number depends on door size, required views, cycle time and sensor path. A system may use one robot, two robots working on opposite sides or a combination of robots and fixed cameras.

Can one system inspect several door models? +

Yes. Product identification can automatically select the correct robot path, camera settings, AI model and tolerance recipe. The cell must be mechanically designed to accommodate every required variant.

Can Intelgic inspect both the inner and outer door surfaces? +

Yes. Multiple robot paths, fixed cameras or separate stations can inspect both surfaces. Deep cavities may require compact or side-view imaging.

Can the system inspect gaps and flushness after vehicle assembly? +

Yes. A 3D sensor can measure door gaps, flushness and feature-line alignment relative to adjacent vehicle panels.

Does AI detect defects it has never seen? +

Anomaly-detection models can flag unusual regions, but every critical defect class and minimum size should still be validated with representative samples before production release.

How are detected defects communicated to repair operators? +

The system can display the defect on a door map, save an annotated image, send the location to a rework station or interface with an automated marking system.

What is required for a feasibility study? +

Intelgic typically needs representative good and defective doors or panels, defect definitions, minimum defect sizes, door variants, line cycle, paint colors, available installation space and acceptance criteria.

16 · Talk to an Intelgic Automotive Inspection Expert

Talk to an Intelgic Automotive Inspection Expert

Contact Intelgic to discuss your car-door manufacturing stage, line cycle, paint finishes, critical defect classes and measurement requirements. Our team can evaluate representative parts and develop an inspection architecture using robotic imaging, controlled lighting, 3D measurement and AI-powered defect detection.

Request a Demo

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.