Knowledge
A vision system starts with the required process change. This may be less manual inspection, fewer rejects, a shorter cycle or fewer operator interventions. Use this knowledge base to define the technical decision and prepare a representative trial.
Start with the process
Choose the question closest to your production problem. The linked guides explain the technical limits and the data required for a feasibility test.
Define which feature must be measurable and how rejects are handled.
Compare fixed mechanics, 2D guidance, 3D guidance and bin picking.
Define code quality, speed, surface and the required system response.
Design for diagnostics, recipe management, maintenance and product changes.
Inspection and identification
Choose by product variation, examples, explainability and validation.
Translate the smallest relevant feature into pixels and optical space.
Limit motion blur and time each image to the product position.
Make the feature under inspection visible across product variation.
Select the reader and lighting for printed and directly marked codes.
Test the mark, surface, viewing angle and process variation together.
Camera and imaging
Camera, lens, lighting, mounting and timing jointly determine whether the relevant feature remains visible. A camera specification alone is not a sufficient selection basis.
Work through ten engineering questions and compare SC1000 to SC6000 from the inspection requirements.
Work from inspection detail, speed, interface and production environment.
Compare continuous material flows with discrete image acquisition.
See how focal length, aperture and distortion affect the image.
Use direction, colour and geometry to create contrast.
Compare principles by surface, range, movement and environment.
Build a shortlist and test it with representative parts.
Robot vision and bin picking
Choose the simplest method that provides the required part position.
Read how sensing, camera selection, grasp planning and recovery affect a bin-picking cell.
Compare feeding mechanics, 2D, 3D guidance and full bin picking.
Calculate with successful picks, retries, recovery and bin changes.
Express position and orientation in the robot coordinate system.
Test stability, reachability, collision-free clearance and release.
Connect camera measurements to robot coordinates and verify the error chain.
Requirements and acceptance
Use data from your own process: product volume per shift, labour minutes, current rejects and rework, downtime, changeover time, maintenance and integration cost. Compare these inputs with the measured trial results. This keeps the payback calculation traceable.
Guides and practical tests
Specify what must be seen, measured or picked and how an error affects production.
Include conforming and non-conforming products and normal variation from different shifts or batches.
Vary position, speed, surface, ambient light and contamination within realistic limits.
Record edge cases, cycle time, false rejects, missed defects and required operator interventions.
