THANK YOU FOR SUBSCRIBING
A picture is worth more than a thousand words. This time-tested maxim illustrates how effective the use of visual explanations can be for conveying complex ideas. It has become increasingly relevant in the manufacturing environment, as the adoption of digital technologies has accelerated. Manufacturers are turning to digital technologies like Industrial Internet of Things (IIoT) applications to paint a picture of their operations to drive improvements in productivity and quality using data collected from their factory floor. Over the next decade, manufacturers are estimated to capture approximately $4 trillion of value from IIoT applications through increased revenues and reduced costs. This value is not easy to realize when the problems are complex and require the observation of countless variables. When traditional sensor technologies fall short of the mark, manufacturers turn to machine vision technology to literally capture a picture that can tackle these problems.
Machine Vision Explained
At their foundation, machine vision systems consist of a camera and a computer or a microcontroller. An image is captured and depending on its quality, may undergo a series of pre-processing events like rotation, brightness enhancement, and image restoration to enable further analysis. Then, software toolsets are employed for use case-specific processing like segmentation and feature recognition to create actionable data points. These datasets are interpreted by analytics tools to inform decisions and trigger programmable logic controllers (PLCs), dashboards, and robots to perform actions like rejecting a part or displaying a warning.
"When traditional sensor technologies fall short of the mark, manufacturers turn to machine vision technology to literally capture a picture that can tackle these problems"
The potential of these systems is vast, as the principal limiting factor to their use is that the key insights they are deployed to find must exist somewhere in the camera’s line of sight. Given the widespread applicability, the unique value of these systems is not in the ability to capture an image, but rather in the algorithms that process those images for a more transparent view of the factory floor.
A Vision Of Quality And Authenticity
One of the most common applications of machine vision systems is in quality assurance and control. Before the digital transformation in manufacturing, trained operators manually inspected parts on the line for defects and compliance. Now, companies are making investments in machine vision technology to automate these activities. The investments are paying off substantially, as these systems can inspect more than 1000 parts per minute with consistent performance and the use of minimal floor space. According to Cognex, a leading supplier of machine vision systems, an automotive client used vision systems to automate inspection operations and increased overall efficiency by 20 percent. Furthermore, this technology creates additional strategic value by capturing inspection data such as rejection reasons and time stamps, so that engineers can quickly isolate the root cause of part defects, and develops solutions to mitigate them.
For larger OEMs that procure high-value parts, machine vision has also been successful in ensuring the quality of the parts they receive from their suppliers. Counterfeit mitigation remains a top priority for these manufacturers and machine vision offerings are proving beneficial in assuring the authenticity of their supply chains.
Old Machines, New Vision
There is no shortage of applications for machine vision technology in the factory. Quality control represents a use case with proven return on investment but more obscure applications, like digitizing legacy machines, are also generating value. Recent estimates indicate that almost 90 percent of machines in factories throughout the world are not connected. These legacy platforms contain many artifacts like gauges, dials, and displays that provide operators with key insights on productivity, quality and machine performance. Unfortunately, this knowledge resides with a single operator and cannot be utilized in any downstream applications unless it is digitized.
Replacing these machines with their newer, smarter counterparts is unrealistic for most companies due to financial and operational constraints but retrofitting them with new technology is a low-cost solution for enhancing the capabilities of these trusted assets. The artifacts discussed above are interpreted based on visual cues, which make them a prime candidate for machine vision retrofit kits. The Digital Manufacturing and Design Innovation Institute (DMDII) has invested in a project that is piloting the development and implementation of these retrofit kits at multiple production facilities. The University of Cincinnati is leading the project with technical support from International TechneGroup Incorporated (ITI) and TechSolve, Inc. Raytheon and Faurecia have guided the development towards real industry use cases and are serving as the pilot implementation sites for the retrofit kit solutions. The kits are non-invasive, can be trained to interpret a variety of different legacy artifacts and can be adapted to fit different budgets.
Visions Of The Future
There is no question that machine vision technology will continue to propagate onto factory floors throughout the United States. As this acceleration continues, it is important that organizations across the manufacturing base think progressively and attempt to address the challenges of an increasingly connected industry. Standardization, cyber security, and accessibility will be important topics to address to aid the effective adoption of this technology. These challenges are very complex and cannot be solved through the efforts of one organization alone. Collaboration among multiple organizations will be vital to see our way to a digitally enabled future.