Machine vision for robots and factory automation

In the beginning it was computer vision, also understood as artificial vision, which is the discipline that studies how to enable computers to understand and interpret visual information found in images or videos. There is a computer that processes images in a real scene, captured through one or more cameras, with the goal of obtaining useful information that enables decisions or actions to be made automatically or semi-automatically.

Research has made enormous strides in computer vision, just think of the many numerous applications in industry.

This has obviously made production much more efficient, minimizing all kinds of waste and eliminating so-called downtime, so the basis of machine vision is the desire to simplify and make production faster and more accurate.

But what are we talking about if we refer to machine vision for robots?

Machine vision for robots is the ability on the part of a robot to reliably detect all kinds of objects to precisely define their dimensions and to avoid any risk of damage.

In addition, machine vision also allows for the detection of anomalies in a decidedly very quick time. Machines have the possibility to be more efficient, this allows companies to achieve results that will be more and more ambitious, in this way we go to reduce the costs needed for productions as well as time.

With machine vision, it can be said that the future therefore seems to be within reach.

Where can we apply machine vision for robots?

The market today has an extreme need for systems that are fast and accurate at the same time, that are versatile and ready to embrace change. Rigid systems have been replaced by mechanisms capable of quickly changing production types without any problem, this system allows for sensitive time optimization.

An example we can give is that of processing and handling of both products and semi-finished goods where the dimensions and positioning are highly variable.

It should not be underestimated that the demand for robots capable of analyzing various scenes is really high. The market when it comes to machine vision has given the idea of not creating robots that are too complex to use or difficult to systematize. This is because the market is pushing for vision that is not too difficult to implement with existing systems.

What is a machine vision system capable of doing for robots?

It can inspect arbitrarily shaped components, freeing the potential of machine vision from the spatial constraints to which it is subject in normalcy. Robotic inspection improves the quality of production, as robots can perform inspections on every part, not just limited to sampling.

One of the interesting topics in Industry 4.0 is that of adapting the behavior of an industrial line in real time, to the point of being able to seamlessly manage a production of unique parts, where each one requires different treatment from all the others. Therefore, machine vision systems can dynamically adapt to each and every part that passes on the line, then inspect all of its characteristics without compromise.

What happens when an automated system has to replace the human being?

If an automated system is to replace the human being in tasks such as “aesthetic” inspection of parts, it must take over as much for the eyes as for the hands. The parts to be screened have geometric shapes that are complex and require observation from different points of view. A machine vision system for robots can process in a short time, so many images from each individual part, thus reaching the human possibilities, exceeding the results, thanks to features such as speed, continuity and objectivity.

If the areas to be inspected require optical homogeneity for each point in the acquired image, linear cameras mounted on the robots allow surfaces to be photographed as if they were on a single plane. The same approach can be used for three-dimensional profile acquisition, with the integration of robots that have 3D displacement sensors.

At first, a 2D or 3D vision system, recognizes shape and position so that an initial robot can pick up the object correctly. The robot then moves the part close to a second vision system so it can inspect it over its entire three-dimensional shape. After that, the part is passed to a second robot to detect the parts hidden by the first gripper. After the inspection is finished, the system figures out the quality of the part and then the second robot puts it in the correct output position.

MACHINE VISION FOR FACTORY AUTOMATION

When we talk about automation in the context of manufacturing processes of industrial nature, it is impossible not to talk about machine vision.

Machine vision in industrial automation is the only solution to optimize productivity and greatly reduce all kinds of production costs, while always ensuring the highest quality.

It should be pointed out that machine vision is also applied in the quality control of every product. Scanners, digital cameras, sensors, image processors, thermal cameras: these, however, are just some of the instrumentations that resort to the use of vision, all of which serve to meet the needs of Industry 4.0!

In every vision system that is used to automate industrial processes, the foundation must be laid on software that will be responsible for processing all kinds of images and transforming what is stored into data.

Machine vision is a virtually essential element of automation systems; there is no other aspect of the production line that can capture more information, evaluate products or search for defects. Vision systems generate large amounts of image data that go on to intensify their usefulness in Industry 4.0.

During the advancement in data analysis, the high volumes of data accessible through vision equipment will be used to identify and mark defective products, understanding their deficiencies, and this will enable fast and effective intervention.

If a machine vision system for factory automation is configured correctly, it will minimize or even eliminate any kind of defect. Machine vision offers the ability to see and identify objects like the human eye, but with greater speed and accuracy. Systems incorporating vision can examine a huge number of parts in seconds and provide reliable and accurate inspection results despite hours of repetitive operations. Aided by high-resolution cameras and highly sophisticated optics, machine vision systems can distinguish even minute differences too small to be detected by humans.

The ability to make a virtual archive of all the data collected through machine vision is really an important step in being able to make almost no time loss to get to the point of reducing production costs. All this will lead to the optimization of time and methods, this means improving efficiency and productivity by empowering personnel to be able to properly execute production processes, in this way all waste will be eliminated.