Scenario

You have a flexible manufacturing process which performs limited runs of similar, but different products, such as logo printing, engraving, or custom part fabrication. 

You also require quality control on this process, as each of the products in a single run need to be the same (or similar within a certain variance). You may currently be performing this quality control with a manual process, where workers view the product as it passes them, and make a judgement as to whether that product is satisfactory or not.

Issue

Using manual quality control introduces human error into the process, as each judgement is based on the worker's assessment, as well as their level of attention.

It would be ideal to replace the manual quality control with an automatic vision process, however this type of system when the product is always the same, and not for the limited run situation.

Specifically, you have investigated Omron's vision capabilities, and found that the FH vision system uses a tool named ‘Shape Search III’, which uses edge detection alongside a taught model of a ‘good’ product to assess a presented product for judgement. However, as it uses a taught model, this tool is normally used for the same product, and if it were to be changed, normally requires a skilled engineer to modify the FH vision system's program.

Solution

The taught model of the Shape Search III tool can be retaught to a limited degree, such that an operator can place a ‘good' product in the camera's field of vision, trigger a signal, and then continue the production run.

Be aware that this process is not as precise as changing the taught model using the FH vision system's program, as the newly taught model cannot be checked or tested before being put into operation. It is also not recommended for high-precision roles.

The purpose of this ‘on-the-fly’ reteaching method is to provide a quick general similarity test, and as such the Shape Search III tool should be configured with a generous margin of acceptability.

In order to reteach the taught model, use the following FH vision flow:

This flow consists of:

2.Set Unit Figure

This flow item is normally used to adjust flow item parameters. In this flow, it is being used to set the opposing corners of the rectangle used to teach the model in the Shape Search III tool.

It was configured as per the below image.

In this case, the two corners are defined as (0,0) and (625,475). These values may vary depending on the camera used, but can be determined through inspection.

By modifying the values of these corners (or redefining the Shape Search III rectangle using the same values), the taught model is forcibly retaught based on the new rectangle and current image.