Crankshaft Surface Defect Inspection: Full-Coverage Vision Inspection of Main and Rod Journals with Borescope Detection

Overview

Crankshafts are complex rotating components with multiple journals, counterweights, flange areas, keyways and internal holes. A reliable inspection system must therefore cover not only visible surface defects, but also deep or internal areas that cannot be inspected effectively with conventional external cameras.

Big Bird Industrial's crankshaft vision inspection solution combines multiple area-scan cameras, customized borescopes, robotic positioning, deep-learning-based machine vision and code traceability. The system is designed to inspect crankshaft surface defects, verify counterweight lightening-hole penetration, detect foreign objects inside center holes, and read product identification codes.

For the referenced application, the published inspection cycle is 48 seconds, with the actual takt configurable according to the workpiece and inspection requirements.

1. Why Crankshaft Inspection Is Challenging

The crankshaft is a core powertrain component responsible for converting reciprocating piston motion into rotational output. Its geometry is highly three-dimensional, with main journals and rod journals distributed along the shaft and connected by crank webs and counterweights.

Depending on the crankshaft design, inspection may cover:

• Main journals
• Rod journals
• Long journals
• Flange journals
• Flange surfaces
• Keyways
• Center holes
• Counterweight lightening holes
• Small end faces

Although many of these areas have similar metallic surfaces, the potential defects can vary significantly. Typical inspection targets include:

• Impact marks
• Scratches
• Spots
• Abrasion marks
• Inclusions
• Cracks
• Black scale or black-skin areas
• Material shortage
• Oil or water contamination

The challenge becomes greater because a crankshaft is a rotating component with multiple inspection angles. A conventional fixed-camera station may not provide sufficient coverage of all required surfaces. The inspection system therefore needs coordinated movement and multi-angle image acquisition.

Counterweight Lightening-Hole Side Penetration

One particularly important inspection item is side penetration of counterweight lightening holes.

Lightening holes are machined to adjust crankshaft balance and reduce unnecessary mass. If a hole breaks through into an adjacent surface or wall, it may affect local structural integrity and create a potential stress-concentration area.

Because this defect can be difficult to identify from an external view, inspection from inside the hole using a customized borescope can provide a more suitable detection method.

2. Inspection Solution: Multi-Camera Vision + Customized Borescopes

The crankshaft inspection system uses multiple imaging methods to address both external surfaces and difficult internal areas.

The published inspection scope includes:

QR Code Integrity

The system checks whether the crankshaft's QR code is complete and readable, providing a basis for product identification and downstream traceability.

Counterweight Lightening-Hole Side Penetration

Customized borescopes are used to inspect the internal geometry of lightening holes and identify abnormal penetration into adjacent areas.

Center-Hole Foreign Object Detection

The internal condition of the crankshaft center hole can be inspected for foreign objects or contamination that may not be visible from an external camera.

Small End Face Defect Inspection

The small end faces of long journals can be inspected for:

• Impact marks
• Scratches
• Spots
• Abrasion
• Inclusions
• Cracks
• Black scale
• Material shortage
• Oil and water contamination

Flange Inspection

The flange area can also be included in the vision inspection program to identify specified surface defects.

Journal and Keyway Inspection

The system can inspect the surfaces of:

• Main journals
• Rod journals
• Long journals
• Flange journals
• Keyways

This multi-area approach allows the inspection program to be configured according to the actual crankshaft geometry and customer-defined acceptance criteria.

3. Camera and Hardware Configuration

The referenced system uses a combination of high-resolution area-scan cameras and customized borescopes.

Imaging System

• 5 MP area-scan cameras: 2 units
• 20 MP area-scan cameras: 12 units
• Customized borescopes: 3 sets
• Large-area light sources: 1 set
• Bar lights: 2 sets

The external cameras provide high-resolution imaging of the visible crankshaft surfaces, while the borescopes extend the inspection capability into areas such as center holes and lightening holes.

Vision and Control

The system integrates:

• QR/code reading
• Machine vision inspection
• Deep-learning algorithms
• Inspection traceability
• Inspection data analysis
• User-defined inspection regions
• Remote inspection viewing

The control system uses two industrial PCs from Advantech (AAEON) according to the published equipment configuration.

Robotic Positioning

Collaborative robot configurations can use brands including:

• Elite Robots
• SIASUN
• Han's Robot
• JAKA

The robot moves the imaging system or coordinates the inspection position according to the programmed inspection path.

4. Technical Specifications

The following specifications are based on the publicly available product information for the referenced inspection system.

The 48-second cycle is a published application value, rather than a universal cycle-time guarantee. Actual cycle time should be calculated according to the crankshaft geometry, number of inspection points, camera configuration, robot path and required inspection criteria.

5. Key Considerations When Selecting a Crankshaft Vision Inspection System

5.1 Define Inspection Items by Location

Crankshafts contain many inspection areas, and different areas may have different defect criteria.

Before equipment design, the inspection specification should clearly define:

• Which surfaces must be inspected
• Which defect types must be detected
• Acceptable and unacceptable conditions
• Required image resolution
• Whether traceability is required

This reduces repeated algorithm adjustment during commissioning.

5.2 Evaluate Deep-Hole Accessibility

For center-hole foreign-object inspection and lightening-hole side-penetration inspection, the key question is not simply camera resolution.

The project should evaluate:

• Hole diameter
• Hole depth
• Internal geometry
• Borescope accessibility
• Required viewing angle
• Lighting conditions

A customized borescope configuration may be required when the inspection area cannot be directly observed by an external camera.

5.3 Treat Code Reading as a Separate Performance Indicator

QR-code inspection is directly connected to product traceability.

Therefore, code reading should be validated independently from defect detection, including:

• Reading success rate
• Code location
• Surface condition
• Contrast
• Marking quality
• Relationship between code data and inspection results

5.4 Validate the 48-Second Takt

Inspecting many areas within one cycle requires parallel image acquisition and optimized robot movement.

During system integration, the cycle-time calculation should include:

1. Workpiece loading
2. Positioning
3. Robot movement
4. Image acquisition
5. Algorithm processing
6. Code reading
7. Result output
8. Workpiece unloading

The published 48-second cycle should therefore be treated as a reference for a specific application rather than a fixed value for every crankshaft.

6. Deep-Learning Vision for Multiple Crankshaft Defect Types

Crankshaft inspection involves defect categories with significantly different visual characteristics. A single conventional rule-based image-processing method may not be sufficient for every inspection item.

The referenced solution uses deep-learning-based machine vision together with inspection regions defined according to different crankshaft areas.

For practical projects, sufficient defect samples are important for both training and validation.

This is particularly relevant for defects such as:

• Inclusions
• Black scale
• Surface scratches
• Impact marks
• Abrasion
• Material shortage

Sample preparation is often one of the most important stages of a machine vision project. The system needs representative examples of both OK products and different types of NG conditions to establish reliable inspection criteria.

7. Big Bird Industrial's Machine Vision Inspection Capabilities

Harbin Big Bird Industrial Co., Ltd. (Big Bird Industrial), formerly Harbin Shimada Big Bird Industrial Co., Ltd., provides integrated solutions covering precision industrial cleaning and machine vision inspection for automotive and precision-manufacturing applications.

Its machine vision business is organized around specialized inspection areas, including:

• Orient See — dimensional and geometric inspection
• Surface See — surface defect inspection
• Assembly See — assembly inspection and poka-yoke
• Inner See — internal-hole and inner-wall inspection
• Paint See — paint-surface defect inspection
• Algorithm & Software Platform — vision algorithms and software systems

For crankshaft applications, Surface See can address external surface defect inspection, while Inner See can be applied to inspection requirements involving internal holes and difficult-to-access areas.

This combination allows different inspection technologies to be integrated according to the geometry and quality requirements of the powertrain component.

FAQ

Why does counterweight lightening-hole side penetration need to be detected?

Lightening holes are used to adjust crankshaft balance and reduce unnecessary mass. If machining causes a hole to penetrate into an adjacent wall or surface, the local structure may be affected and the area may become a potential stress-concentration location.

Because the defect may not be visible from outside the hole, a borescope can be used to inspect the internal area.

Can one vision algorithm detect all types of crankshaft surface defects?

Not necessarily.

Crankshaft defects can differ considerably in shape, texture, contrast and location. Deep-learning algorithms can be configured for different defect categories and inspection regions, but reliable performance depends heavily on representative training and validation samples.

Defects such as inclusions and black scale can be particularly dependent on material and surface characteristics, making sample collection an important part of the project.

Can 48 seconds really cover so many inspection areas?

The referenced system uses multiple cameras operating in parallel, with different cameras assigned to different inspection regions. Robotic positioning can also move the imaging system to required inspection locations.

The published 48-second cycle applies to the referenced application. Actual takt time should be calculated from the specific crankshaft, inspection coverage, robot path and image-processing requirements.

Can oil or water contamination be mistaken for a surface defect?

Yes.

Oil and water can create reflections, dark areas or other visual patterns that may resemble certain surface defects. They are therefore listed as specific inspection targets in the referenced system.

Image algorithms can use characteristics such as shape and location to distinguish different conditions. However, if excessive oil or water remains on the workpiece, upstream blow-off or cleaning can help reduce visual interference before inspection.

Conclusion

Crankshaft vision inspection is not simply a matter of installing a high-resolution camera. The complex geometry of crankshafts requires a combination of multi-angle imaging, robotic positioning, customized borescopes, deep-learning algorithms and traceability functions.

For external journals, flange surfaces and keyways, multi-camera vision provides broad surface coverage. For center holes and counterweight lightening holes, borescope inspection extends machine vision into areas that conventional external cameras cannot directly observe.

For the referenced application, the system provides a 48-second published inspection cycle, with the final configuration and takt determined by the specific crankshaft geometry and inspection requirements.