Machine vision is how automation systems see. At Motionwell, we integrate vision systems into production lines to perform inspection, measurement, guidance, and verification tasks that would otherwise depend on human operators working under magnification or with gauges.
We work with Keyence, Cognex, and Basler camera platforms depending on the application requirements. Our vision integration covers everything from simple presence/absence checks to multi-camera dimensional measurement and AI-assisted defect classification.
Four Jobs Vision Does in a Production Line
Every vision application we deliver falls into one of four categories. Understanding which category your application belongs to determines the camera, lighting, and algorithm approach.
1. Defect Detection
The camera captures an image of every part at line speed and compares it against trained references. The system makes a binary decision — pass or fail — and triggers a reject mechanism for failures.
What matters here is not the camera resolution. It is the lighting. The same scratch that is invisible under diffuse lighting becomes obvious under directional or dark-field illumination. We spend the most engineering time on lighting geometry because that is where most vision projects fail.
Our approach: We test your actual parts under 4-6 different lighting configurations before specifying a system. If the defect is not reliably visible under controlled conditions in our lab, we will tell you before you spend money on a system that disappoints on the production floor.
2. Dimensional Measurement
High-resolution cameras with calibrated optics measure critical dimensions inline — pin positions, gap widths, component heights, alignment offsets. Results feed directly into SPC systems for trend monitoring.
The challenge is not accuracy (modern cameras easily resolve to sub-pixel). It is repeatability across thermal drift, vibration, and part variation. We design measurement cells with thermal compensation, anti-vibration mounting, and calibration routines that your operators can run without an engineer present.
3. Vision-Guided Alignment
Robots need eyes. When parts arrive with variable position and orientation — from conveyors, feeders, or manual loading — vision tells the robot exactly where to pick and how to place.
This is where camera-robot calibration matters most. We use hand-eye calibration procedures that achieve repeatability within ±0.02mm, validated with statistical capability studies on your actual parts.
4. Label and Code Verification
Every product leaving the line must carry correct, readable identification. We integrate print quality inspection, barcode/QR readability grading (per ISO/IEC 15416 and 15415), and OCR/OCV verification for date codes, batch numbers, and serial numbers.
For pharmaceutical lines, we add red strobe inspection for serialization verification — catching print defects that are invisible under normal lighting but cause read failures at distribution.
How We Select Vision Hardware
Camera and lens selection follows a systematic process, not brand preference:
- Keyence IV3 series — compact, self-contained inspection for simple pass/fail applications. Fast setup, limited customization.
- Keyence CV-X series — multi-camera controller for complex inspection routines with up to 4 cameras per controller
- Cognex In-Sight — deep learning capable for defect classification where rule-based algorithms struggle
- Cognex DataMan — barcode and code reading with high-speed decode and grading capabilities
- Basler industrial cameras — GigE and USB3 cameras with custom optics for high-resolution measurement applications
- Ring lights — uniform illumination for flat surface inspection (most common starting point)
- Coaxial lighting — reveals surface defects on reflective or polished parts
- Backlight — silhouette measurement for dimensional accuracy and edge detection
- Dome lighting — diffuse, shadow-free illumination for curved surfaces and complex geometry
- Structured light — 3D surface profiling for height measurement and flatness inspection
Our Integration Process
Every vision project starts with a sample study. This is non-negotiable — we will not spec a system without testing your actual parts first.
- Sample study and feasibility. We photograph your parts under multiple lighting conditions, test detection algorithms, and report achievable accuracy and speed. If it cannot be done reliably, we tell you here.
- Optical design. Camera resolution, lens working distance, field of view, and lighting geometry are specified to match your inspection requirements and physical constraints.
- Algorithm development. We configure inspection routines using the camera platform’s native tools, avoiding custom software when possible for maintainability. Your operators should be able to adjust thresholds without calling us.
- PLC integration. Vision results drive sorting, rejection, or process adjustment in real time. Every result is logged with timestamp, station ID, and pass/fail status.
- Capability validation. We run GR&R studies on your production parts to confirm detection rates, false positive rates, and measurement repeatability before handover.
When Vision Is the Right Answer (and When It Is Not)
Machine vision adds the most value when:
- Manual inspection is a bottleneck or source of inconsistency
- Defects are escaping to downstream operations or to customers
- You need measurement data for SPC or process improvement
- Part positioning varies and robots need guidance
- Regulatory requirements demand 100% inspection with documented results
Vision is not the right answer when the defect has no visual signature (internal voids, chemical composition), when the inspection speed requirement exceeds camera frame rates, or when environmental conditions (steam, oil mist, extreme vibration) make reliable imaging impractical. In those cases, we recommend alternative sensing approaches.
If you are unsure whether vision is the right approach for your application, send us sample parts and a description of what you need to detect. We will give you an honest assessment of feasibility before any commitment.