Pick and place vision: what to assess
When vision contributes, when good feeding is enough and what must be solved before choosing camera and robot.
Vision in pick and place does not start with the camera: it starts with the part and how it arrives. If position varies, you must measure; if the part can always be presented the same way, a tray or fixture may be more robust. The decision uses real variation and a trial.
What vision adds
Vision locates the part, orients it, verifies presence and can guide picking or inspection. It does not improve feeding that cannot separate parts or a grip that cannot handle the geometry.
2D, 3D or bin picking
The choice depends on the missing information. The table organizes the case.
| Criterion | What it solves | Limit |
|---|---|---|
| 2D vision | Planar position and presence. | Does not solve height or spatial orientation. |
| 3D vision | Height, volume and orientation. | More calibration, data and cost. |
| Bin picking | Stacked or random parts. | Fails with tangles, transparency or low contrast. |
Lighting and calibration
A camera does not compensate for bad lighting. Contrast and reflection define whether the part is visible. Robot-camera calibration maps pixels to robot coordinates; without it, vision locates but the robot does not reach.
Gripper and feeding
Vision does not replace the gripper. Vacuum, parallel gripper, suction cups or soft gripping must be defined, along with how parts are presented. Feeding remains part of the cycle and dead time.
Conveyor tracking and PLC
If the part moves, an encoder, synchronization and logic that provides position to the robot are needed. You must also define what happens if the part is not found or is lost.
When vision is not needed
When the part arrives in a fixed position, fixturing always presents it the same way or a better tray removes variation. In those cases vision adds cost and maintenance.
How to validate
With real samples, site lighting, good and bad parts and an acceptance criterion. A demo with the perfect part does not replace production testing.
Frequently asked questions
When is vision worth it in pick and place?
When position varies, sorting or inspection is needed, or the part moves. If presentation is fixed and repeatable, good feeding may be enough.
2D or 3D?
2D for planar position and presence; 3D for height, volume or spatial orientation. The decision is validated with real samples.
How is the camera calibrated to the robot?
With a procedure that maps pixels to robot coordinates and checks error at several points. Without calibration, vision locates but the robot does not reach.
Does vision replace the gripper?
No. The gripper or vacuum cup defines whether the part can be picked. Vision only helps find and orient it.
What happens with parts that cannot be seen?
A rule is defined: retry, reject or request intervention. The cell must not get stuck waiting for a decision nobody made.
How is it validated?
With real parts, site lighting, position variation and an acceptance criterion. The trial must include difficult parts, not only the perfect sample.
