How a 3D profiler measures
It projects a laser line onto the part and a camera views it from another angle. From how the line is distorted it calculates the height of every point. As the part passes underneath, or the robot moves it past, a point cloud of its full shape is built up.
That cloud is compared with a good part or its CAD. Wherever the surface deviates more than allowed, there is a defect.
Which defects it detects
- Wrinkles. Waviness on flanges and flat areas that can barely be seen in a flat image.
- Splits and necking. Tears or thinning in radii and deep-drawn areas.
- Out-of-tolerance dimensions. Flange heights, angles and hole positions.
- Missing operations. Unpunched holes or unbent flanges.
2D or 3D
| Criterion | 2D vision | 3D vision |
|---|---|---|
| What it sees | The part as a photo | The shape and the heights |
| What it is for | Position, presence, codes and holes | Wrinkles, deformation and height dimensions |
| What affects it | Lighting, glare and oil on the sheet | Very shiny surfaces and the viewing angle |
| Typical use on press lines | Vision-guided loader and code reading | Inspecting the part at the line exit |
Many lines use both: 2D to guide the robot at the loader and 3D to inspect the part at the exit.
How it is integrated into the line
- Where it goes. Over the exit conveyor with the part moving, or at a station where the robot moves the part past the sensor.
- Trigger and travel. The sensor needs to know when the part arrives and how fast it moves: a PLC signal or an encoder on the conveyor.
- Result to the PLC. Good or bad part on every cycle. Bad parts are separated without stopping the line.
- Logging. Every result is saved with its part number and time, and can go to the MES over MQTT.
- Maintenance. Cleaning the sensor window, checking with a master part and backing up the configuration.
What to bear in mind
- Oiled or galvanised sheet is shiny: the sensor and angle must be chosen for the real surface and tested with production parts.
- The smaller the defect, the smaller the area each sensor covers: sometimes several are needed.
- The part must arrive in a repeatable position, or the program must correct its position before comparing.
- You need to agree on what counts as a defect: tolerances and examples of good and bad parts.
What we need to propose the inspection
- The part or its CAD, and the defect or dimension you want to check.
- How fast the part moves.
- Where the camera would go on the line.
- What to do with bad parts.
- Good parts and defective parts for testing.
Send us the layout and the part. The first proposal is free of charge.
Frequently asked questions
What is the difference between 2D and 3D vision?
2D sees the part like a photo: position, presence and codes. 3D measures height, so it detects wrinkles, deformation and dimensions that a flat image cannot show.
Can parts be inspected without stopping the line?
Yes. The profiler measures the part as it passes on the conveyor or while the robot moves it.
Can you add vision to an existing cell?
Yes. We integrate the camera with the robot and PLC you already have.
Which equipment do you work with?
We integrate Cognex cameras and LMI Gocator profilers. If your plant has a different vision standard, we work within it.