As an Intel® Gold member for IoT, we at DnA Industry Solutions have developed a Market Ready Solution for the digital and analytics implementation of AI-powered machine vision for Industrial operations. Deploying IoT solutions in manufacturing allows for increased connectivity between physical manufacturing objects whereas the use of AI-powered solutions enables the devices to learn from the data and automatically make necessary adjustments in the manufacturing process. Encompassing both IoT and AI technologies, our solution can benefit manufacturers in a wide range of industrial sectors, allowing them to unlock significant value gains across different manufacturing functions.
While manufacturers are increasingly adopting intelligent technologies and AI-powered systems to increase the efficiency of their processes, significant challenges remain, thus drastically capping the value capture from such solutions. Manufacturers are struggling to keep up with rapidly advancing technologies and are unable to find solutions specifically applicable to their needs. Furthermore, solutions may be implemented in siloes, or may not be well suited for the individual circumstances and needs of the manufacturing plant. The lacking in-house expertise is also a significant factor hindering the adoption of intelligent solutions.
AI-powered machine vision systems are highly value-adding and efficiency-increasing in particular for controlling the production quality, where the changes affecting the quality may be difficult or impossible to detect by human eye. Leveraging Intel® technology, our user-friendly AI vision system allows for a wide range of process data as an input. Using data collected by industrial cameras, the AI system can be trained for automated quality control. This greatly reduces the need for human intervention, makes it possible to detect issues in an earlier stage of production and thus overall increases the efficiency and accuracy of the quality control process. Our implementation solution offers all that a manufacturing plant needs to improve its efficiency in terms of flexibility, scalability, security, and data protection.
Elastic wristband production using AI powered vision system
Our market ready solution has been successfully implemented in a range of manufacturing settings. A German model textile manufacturing plant producing elastic wristbands saw excellent results after deploying the AI-powered machine vision system in the textile industry, where very small changes in the yard tension or its physical characteristics can drastically affect the output quality. Such changes are almost impossible to detect by human eye, and because quality issues are only discovered at later stages of production, the time required for correction and rework is high, and the late detection leads to scrap losses.
Leveraging Intel® technologies, our solution increased production efficiency by introducing two parallel functions: a data driven quality inspection system enabling the identification of issues in yarn tension and quality in real time, as well as an AI-based machine vision system observing the printing quality of the wristbands. Thanks to our partnering ISV ANTICIPATE’s software tool, we were able to reduce our usual implementation time by 3 months.
Where such implementation projects usually take around 3 months, we were able to bring this time down to a mere 8 weeks by partnering with ANTICIPATE. Their software tool allowed us to automate the implementation process, leading to significant savings in implementation efforts, thus making our offering more cost-efficient to our clients.
To control the yarn tension and quality, we implemented Intel® Edge Insights Industrial software in conjunction with a neural network using data from yarn tension sensors. This allowed for the automated monitoring of the yarn tension and quality in real time, with the system signalling to the human operators if quality errors were detected to ensure that necessary adjustments could be made during production. This reduced quality control actions by 50 percent and reduced scrap losses by 40 percent.
Additionally, using an Intel® Distribution of OpenVINO toolkit, the AI-driven machine vision application based on machine learning was implemented for quality control and improvement. As opposed to manually coding a set of rules for acceptable and not-acceptable parts, the deep learning neural network was able to automatically analyse a large set of pictures from the wristbands and develop rules to classify them. The system could then recognize the output quality as acceptable or not acceptable, automatically adjust the printer settings, and notify the operators of any issues, allowing faster development time. This solution increased the overall system performance while reducing CPU usage by 50 percent.
Our Market Ready Solution implementing the AI-powered machine vision system can easily be adjusted to the specific needs of the manufacturing setting due to its flexibility. This allows manufacturers in different industrial sectors to effortlessly capture the benefits of smart manufacturing and become a part of Industry 4.0. For any inquiries concerning our solution, do not hesitate to contact us.
Sources: Intel® technologies and AI Analytics Improve the Quality in Manufacturing https://www.intel.com/content/dam/www/central-libraries/us/en/documents/intel-tech-improve-product-quality-via-ai-wp.pdf