1. Software Master Mode

Vision (OCR and Barcodes)

We have basically gone through the Positioning functions of Vision. In this chapter, we look at Inspection. Note that Inspection does not have a Point function like Positioning; it simply performs visual recognition.

Drag in an Inspection node and click its pencil icon to open the Vision editor. In Vision Application Selection, choose AOI Recognition → Next. Under this category, you can recognize text and numbers, which is OCR, read two-dimensional barcodes, read QR Codes, and so on.

In the left flow blocks, use the same setup method as before to set Motion and Camera. After setting them, click under Camera. This enters the Add Vision Flow page. In the Identify category, you can find the functions mentioned earlier. Choose the target you want to recognize. The Additional Functions section can expand all functions. Click OK, and you can start setting that recognition task.

Using Optical Character Recognition 2 as an example, the Model Selection dropdown on the right provides three options:

• Common: numbers (0 to 9), uppercase Latin letters (A to Z), and symbols (/, @, :, (, ), -, ., #, $, %, &, *, +, <, =, >)
• Numeral: numbers (0 to 9), plus - and .
• Universal: numbers (0 to 9), uppercase and lowercase Latin letters (A to Z), and symbols (/, @, :, (, ), -, ., #, $, %, &, *, +, <, =, >)

The recognition result can be found in the variable list.

AI+

AI+ is actually very powerful, and it is almost guaranteed to appear in advanced finals. However, because of the YMSH robot arm environment, we cannot practice this part on the school arms. So in this chapter we only explain the concept. The detailed operation can only be learned in TM Robot's training course.

AI+ is a major category under Vision. It can do classification, object detection, semantic segmentation, anomaly detection, and instance segmentation. The one we use most often is classification. By taking several photos, it can do a simple form of machine learning, teach the machine to classify objects, and then run different flows according to the detection result.

Single-Step Execution

When editing a program, you can click at the top to enable single-step execution. Click the node you want to run, then use PLAY on the controller to execute it one step at a time. However, only some nodes support single-step execution. For example, Set cannot be executed this way.

A small trick: after creating a vision task, if you accidentally move to the workpiece position, or if you want to reset a Point based on Vision Base, the Vision Base position may shift because the workpiece position is different. In that case, use single-step execution to run Vision once. It will use the position seen during that execution, letting you set a new Point accurately without recreating the vision task.