Oxford Industrial Automation
Vision-guided sorting

Find, classify and handle products at production speed.

Combine machine vision with high-speed robot handling to locate randomly presented products, identify orientation or variant and place each item into the correct process or pack.

Production outcomes

What the cell is designed to achieve.

Performance is confirmed against representative products, the complete process and agreed acceptance criteria.

01Sort mixed or randomly positioned products

Application outcome to validate during concept engineering.

02Orient items before packing or assembly

Application outcome to validate during concept engineering.

03Route products by inspection or identification result

Application outcome to validate during concept engineering.

04Reduce dependence on rigid mechanical presentation

Application outcome to validate during concept engineering.

Engineering sequence

How the application is defined before equipment selection.

The concept is developed from the full production sequence. This prevents the robot from being selected around one isolated pick while missing presentation, machine time, product variation or operator interaction.

  • Define the decision the vision system must make
  • Collect representative samples and defects
  • Engineer lighting, optics and field of view
  • Calibrate camera, conveyor and robot coordinates
  • Design gripper and collision rules
  • Validate rate, false rejects and uncertain-result handling
Project information

What to send for a useful feasibility review.

The quality of the first concept depends on the information available. Photographs and video are useful, but product and process data allow payload, reach, cycle and interfaces to be checked more reliably.

  • Product range, colours, finishes and defects
  • Conveyor speed, width and spacing
  • Required pick rate and placement destinations
  • Product overlap or occlusion conditions
  • Inspection resolution and acceptance criteria
  • Ambient light and cleaning environment
Practical answers

Frequently asked questions

These answers provide an initial planning framework. Final requirements are confirmed against the product, process, site and applicable safety obligations.

Can a vision robot pick overlapping products?

Sometimes, particularly with 3D vision, but heavy occlusion can hide valid grip points. Product presentation may still need conditioning or recirculation.

Can the system inspect and sort at the same time?

Yes, if the image quality, resolution and cycle allow. Inspection results can assign each product to a destination or reject route.

What happens if the camera cannot identify a product?

The controls should define an uncertain-result strategy such as reject, recirculate, stop or request operator review.

How is conveyor tracking achieved?

An encoder measures belt movement. The vision and robot systems use that motion to update each product position until pick interception.

Project review

Review this production task with us.

Send the product details, required output, available space and a photo or short video of the current process. We will review the application and identify the next engineering step.

Send your project details →
☎ 01865 416830Project brief →