AI Trend in Smart Manufacturing

The rise of Industrial IoT and Industry 4.0 have had a transformative effect on the manufacturing industries. They offer, by providing precise, up-to-date and comprehensive data, the opportunity to identify hitherto unknown relationships between machines, processes and logistics, allowing users to take action to improve quality, increase efficiency and reduce waste.

And yet, despite the large number of pilot projects and early implementations, many users still struggle to realize the potential benefits at scale. We have become very good at collecting and collating vast amounts of data, but users still have problems with the next phase, which is actually taking action. They have all of this actionable information, but somehow, they still don’t believe the technology and feel they have to engage their senior consultants extensively to figure out what to do next.

AI offers a potential solution to this logjam by providing a virtual co-pilot for manufacturers. By providing real time insights into the collected data, an organization’s paradigm experts are released to focus exclusively on solutions and process improvement rather than performing vast amounts of analysis and theorizing. It offers trusted advice to empower humans to cope with situations where urgency and complexity combine. This helps organizations to take action much more quickly and with greater confidence, which increases agility and with it brings all kind of benefits.

Once the preserve of large corporations, the cost and complexity involved in implementing AI solutions within manufacturing has reduced to the point where it is now within the reach of even the smallest manufacturing organizations. Applications such as visual inspection, predictive maintenance and the use of cobots to assist the human workforce with repetitive or hazardous tasks are all gaining traction as more and more manufacturers understand the competitive advantages they bring.

And yet there are still hurdles to be overcome. Applying AI to problems without including all of the relevant contextual information can lead to misleading results. Manufacturers also need to consider the implications for the roles and responsibilities of staff, and the skillsets that will be needed in the future. This however, also presents opportunities. The manufacturing industries are suffering from an emerging skills gap as its workforce ages. The new roles created by AI, focusing on creativity, problem solving and process improvement, offer much more dynamic and attractive career prospects to Generation X than the old, repetitive, boring tasks which have characterized much of manufacturing in the past.

It’s important to understand that the benefits of AI can be applied just as easily into existing ‘brownfield’ production lines and machinery as they can on new factories. Technology allows seamless data recovery from legacy PLCs, sensors and machine control systems, together with the electronic replacement of offline paper or spreadsheet based operator data, creating a comprehensive data lake upon which the AI can operate.

Equally important is for users to realize that an AI implementation no longer requires factory floor deployments of large, high-end, and expensive computers. While it's true that robust servers are essential for training AI models, the actual AI systems deployed in factories are a different story. These AI inference devices, which leverage trained models without the need to generate them, are implemented using compact, durable, and budget-friendly industrial computers.

In this latest Innotalks session, host Tim Taberner discusses these topics in detail with Willie Lin, Director, Advantech WISE-IoT, Cathy Yeh, Principal PM Group Manager, Industry Solutions Engineering Asia Microsoft Corp and Paul Turner, President and Chief Operating Officer, Raven.AI.

• Innotalks Part 1: AI Current Status
• Innotalks Part 2: Implementing AI in Factories
• Innotalks Part 3: Key AI Applications