How to choose the eye vision system of industrial robots

This class usually contains one or more cameras and lenses (optical components) for taking pictures of the object being inspected. Depending on the application, the camera can be based on the following criteria: black and white, composite color (Y/C), RGB color, non-standard black and white (variable scanning), progressive-scan or line scan.

The lights are used to illuminate the parts so that better images can be taken from the camera, and the lighting system can be in different shapes, sizes and brightness. Typical lighting forms are high frequency fluorescent lamps, LEDs, incandescent lamps, and quartz-halogen fibers.

It usually appears in the form of a grating or a sensor. When the sensor senses that the component is close, it will give a trigger signal. This sensor tells the machine vision system to capture images when the part is in the correct position.

Also known as a video capture card, this component is usually a card that is plugged into a PC.

This capture card connects the camera to the PC. It takes data (analog or digital) from the camera and converts it into information that the PC can process.

It also provides signals that control camera parameters such as trigger, exposure time, shutter speed, and more. There are many forms of image capture cards that support different types of cameras and different computer buses.

Computers are a key component of machine vision.

For applications, the PenUum or higher CPU is usually used. In general, the faster the computer is, the shorter the time it takes for the vision system to process each picture.

Since there are often vibrations, dust, heat radiation, etc. in the manufacturing site, industrial grade computers are generally required.

Machine vision software is used to create and execute programs, process captured image data, and make "PASS/FAIL" decisions.

Machine vision can take many forms (C library, AcTIveX control, click programming environment, etc.), can be a single function (such as design only to detect LCD or BGA, alignment tasks, etc.), or it can be multifunctional (such as design A kit that includes metering, barcode reading, robot navigation, field verification, and more.

Once the system completes this test, this part must be able to communicate with the outside world, such as the need to control the production process, and send "pass/fail (PASS/FAIL)" information to the database. Typically, a digital I/O board and/or a network card are used to communicate the machine vision system with external systems and databases.

Configuring a PC-based machine vision system Careful planning and attention to detail can help you ensure that your inspection system meets your application needs. Here are some points you must consider:

Determining your goals, this may be the most important step in deciding what you need to achieve in this inspection task. Detection tasks are usually divided into the following categories:

Measurement or measurement

Read character or code (barcode) information.

Detecting the state of an object

Cognitive and recognition special feature pattern recognition

Compare or match objects to templates

The navigation process for a machine or robot can include only one operation or multiple tasks related to the inspection task.

In order to confirm your task, you should first be clear about the tests you need to do to maximize the inspection of the components, that is, you can consider the defects that will occur.

In order to clarify which is the most important, it is best to make an evaluation form that lists the tests that must be done and can be done. Once the main test standards are satisfactory, more tests can be added to improve the inspection process. It is important to remember that adding tests will also increase the time of testing.

Determine the speed you need – how long does it take for the system to detect each part?

This is not only determined by the speed of the PC, but also by the speed of the production line.

Many machine visions include a clock/timer, so the time required to detect each step of the operation can be accurately measured. From this data, we can modify our program to meet the time requirements. Typically, a PC-based machine vision system can detect 20-25 parts per second, closely related to the number of parts detected and the speed of the processing program and the computer.

The performance of a machine vision system is closely related to its components. In the selection process, there are many shortcuts that may greatly reduce the efficiency of the system, especially in optical imaging. Here are a few basic principles you must keep in mind when selecting parts.

1. The choice of camera camera is directly related to the needs of the application, usually considering three points:

a) black and white or color;

b) movement of the component/target;

c) Image resolution.

Black and white cameras are mostly used in inspection applications because black and white images provide 90% visual data and are less expensive than color. Color cameras are mainly used in some situations where color images need to be analyzed. Depending on whether the part moves during the test, we decide whether we choose a standard interlaced camera or a progressive scan camera. In addition, the resolution of the image must be high enough to provide enough data for the inspection task. Finally, the camera must be of good quality and avoid the effects of vibration, dust and heat in the industrial field.

2. The critical element of optical components and lighting is often overlooked. When you use a poor optical component or illumination, even if you use the best machine vision system, it does not perform as well as a low-capacity system with good optics and proper illumination. The goal of the optical component is to produce an image of the best and largest available area and to provide the best image resolution. The goal of illumination is to illuminate key features of the part that needs to be measured or detected. Typically, the design of a lighting system is determined by factors such as color, texture, size, shape, reflectivity, and the like.

3. Image Acquisition Card Although the image acquisition card is only a part of the complete machine vision system, it plays a very important role.

The image capture card directly determines the interface of the camera: black and white, color, analog, digital, and so on.

Using an analog input image capture card, the goal is to convert the image captured by the camera to digital data as much as possible. Using an incorrect frame grabber may result in incorrect data.

Industrial image acquisition cards are commonly used for inspection tasks. The multimedia capture card does not need to be used in this field because it changes image data through automatic gain control, edge enhancement, and color enhancement circuitry. The goal of an image capture card using digital input is to convert and deliver the digital image data output by the camera to a PC for processing.

Consider various changes: the human eye and the brain can identify the target under different conditions, but the machine vision system is not so versatile, it can only work according to the tasks programmed. Knowing what your system can see and what you can't see can help you avoid failures (such as thinking good parts are bad) or other detection errors. Common considerations include component color, ambient light, focus, position and orientation of the component, and large variations in background color.

Choosing the right software: Machine vision software is the smart part of the inspection system and the core part. The choice of software determines the time you write the debug test, the performance of the test, and so on.

Machine Vision provides a graphical programming interface (commonly referred to as "Point&Click") that is generally easier than other programming languages ​​(such as Visual C++), but has some limitations when you need some special features or functionality. Code-based software packages, while very difficult and require coding experience, have greater flexibility in writing complex special application detection algorithms. Some machine vision software also provides a graphical and code-based programming environment that provides the best of both features, providing a lot of flexibility to meet different application needs.

Communication and Recording Data: The overall goal of machine vision systems is to achieve quality inspection by distinguishing between good and bad components. In order to achieve this, the system needs to communicate with the production pipeline so that it can do something when it finds bad parts. Usually these actions are through digital I/O boards that are connected to the PLC in the manufacturing line so that the bad parts can be separated from the good parts. Exceptionally, the machine vision system can be connected to the network so that data can be transferred to the database for logging data and for quality controllers to analyze why scrap occurs. Careful consideration in this step will help to seamlessly integrate the machine vision system with the production line. The issues to consider are:

What is the file format that is transmitted over the network? The RS-232 port is usually used to communicate with the database to record data.

Prepare for the future: When you choose parts for your machine vision system, keep in mind the future production needs and possible changes. These will directly affect whether your machine vision hardware and software can be easily changed to meet new tasks in the future. Preparing in advance will not only save you time, but also reduce the price of the entire system by reusing existing inspection tasks in the future. The performance of a machine vision system is determined by the worst part (just like the capacity of a wooden barrel is determined by the shortest piece of wood), and the accuracy is determined by the information it can acquire. By spending time and effort on the system, you can build a zero-fault and flexible vision inspection system.