Visual A.I. for Quality Control
Industry 4.0 adoption is rising with emphasis on mass customization, rapid product cycles, and faster delivery. This opens the scope for technology led advanced manufacturing practices to create high quality digital smart products efficiently. Quality inspection on the shop floor is one such area where technologies like IoT and Artificial Intelligence is gaining traction.
Currently, most manufacturers rely on human quality inspections to detect flaws, such as dents and cracks in the finished products. However, such human inspections are prone to errors owing to constraints posed by speed, accuracy, and cost. On the other hand, there is no way to identify causal factors for quality deviations, such as quality of raw input, operating conditions etc.
How to enable automated quality detection with supplier quality checks to identify the root cause of errors?
Intelligent technology as an enabler
Computer vision is the fastest growing AI field with plenty of research to bring about industrial applicability. See the snapshot below from arXiv.org, an online archive of research articles.
According to Mckinsey, Manufacturing & Risk are the two business functions that have reported significant benefits from deploying computer vision.
Therefore, what we propose is a next generation solution called Visual A.I. for Quality Control specific to the discrete manufacturing. In the first phase, it will optimize the quality inspection process and later extend the value by bringing in supplier material checks through correlation.
The solution combines algorithm-based image analysis with process parameters, such as temperature and pressure to perform quality inspections. Additionally, such a system correlates supplier material quality, such as chromium, nickel composition in the input steel sheets, with output defects to better evaluate the supplier performance.
Snapshot of the visual quality dashboard in the proposed solution
The solution increases the throughput by operating 24/7 in a consistent manner by eliminating human error caused due to speed of manufacturing lines and fatigue due to surrounding noises. Additionally, it decreases scrap by correcting manufacturing tolerances. Finally, the solution makes the supply chain more agile & responsive by correlating faults to input material quality and thus providing greater visibility to origins of defects in the supply chain.
Additional read: For further details into personas and proposed solution architecture, please access this ebook.
I am looking for partners and customers who would be willing to co-innovate. So, if you are interested in collaborating further on this topic, please reach out to me.
Thank you very much for this interesting blog.
I was just wondering if you could give some update on the maturity of the ideas you posed above, and what functionalities you would see on the roadmap for the coming months/years.
Also, do you only see this fit for manufacturing purposes? Or also other lines of business?
Thanks and best regards,
Hello Nordin, thanks for taking the time and your interest. Like I said, at this point – I am looking for customers and partners to co-innovate and build this functionality. In terms of maturity, we have done a PoC on the same concept using our data management suite, including ML. Further, customers in the discrete manufacturing space have shown interest in the idea as well. I’d be happy to discuss this in detail, if required.
Hi Ankit, your email address was removed. Posting it violates our rules of engagement. Please see the rules here for your reference: https://community.sap.com/resources/rules-of-engagement
Best regards, Mynyna
SAP Community Moderator
elunic AG is proud and very happy to leverage this great mission as a SAP Partner with AI.SEE™!
AI makes the difference.