Computer vision technology in the industry environment is shifting in importance, according to our recent survey. 33% of our respondents from Greater China, who are likely to implement computer vision*, continue to place more weight on smart factory automation, compared to only 22% in the rest of the world. This disparity could be explained by the push for automation encouraged by the Made in China 2025 initiative, but it remains unclear, which specific uses the realm of computer vision will have on Industry 4.0.
Factory automation encompasses a whole range of techniques and technologies, including machine vision, image processing, robotic vision, as well as computer vision. This article takes a look into three applications of computer vision that currently have, and will have, a significant impact on manufacturing.
Computer vision extracts relevant information from images to make sense of them. For instance, a computer vision system (CVS) could collect and analyse images to recognize the objects, including their size and color. Once this object detection and these characteristics have been processed, the information can be passed on to other systems to take action.
Action could come from human intervention, but more and more factories are choosing robotics for a faster, safer, and more efficient solution. Some robots even have cameras embedded in their arms or heads to feed the information collected directly back to the robot.
Automated analysis and action, through the collaboration between a CVS and robotics, can open countless improvements and possibilities to industry. Computer vision systems in recycling plants can recognize objects, like glass through detecting their color and shape and, consequently, use robotic machinery to place similar glass types together.
With this advance in technology, people can also work safely with robots in the same space. Robots with embedded cameras can recognize approaching humans and either pause their activity or be sure to work around the humans. No longer do robots need to be confined in glass boxes in the fear that they may be a hazard to the surrounding workers.
Robotics leads into a very popular and active application of computer vision. Quality control is a process that is relatively simple to automate. Unlike robotics, even detecting faulty products, without taking automated action, can save masses of time and greatly improve accuracy. After a CVS recognizes the damaged product, humans can then intervene to remove or correct it.
Because of its large impact, without the more complicated integration with robotics, a CVS designed for quality control is a good starting point for factories looking to transition into automation.
The concept of a digital twin is to create a digital model of a factory, which can be used to optimize operations and design. Sensors from the factory floor and algorithms, based on computer vision techniques, feed near-real-time information into the digital twin. This data can be collected from a range of technologies, such as cameras, laser scanners and radars.
The factory model combined with the near-real-time data makes it possible to test complex design improvements virtually before making physical changes to the manufacturing process. These improvements include sensors on the machinery itself, ensuring apparatus is fixed or replaced before it is nonfunctional. Aside from a greater degree of accuracy and efficiency, a digital twin can also transform the factory into a dynamic, responsive producer.
Robotics, quality control and digital twins are only the beginning as manufacturing starts to transform into an automated smart factory. Apart from computer vision technologies, there are many more changes to come, notably the possibilities offered by the highly connected Industrial Internet of Things (IIoT).
[bs_button size=”md” type=”link” value=”To explore VIA’s computer vision solutions click here” href=”/en/systems/computer-vision-solutions/”]
*These respondents answered “No” to the question “Have you had previous experience working with computer vision technologies?” and preceded to respond with “Very likely”, “Somewhat likely” or “Neither likely or unlikely” to the question “How likely are you or your company to develop computer vision systems in the next 2 years?”.