AI and the manufacturing industry

29 January 2019

In a presentation about the rise of cognitive factories and the use of AI on the plant floor Karthik Sundaram, Frost & Sullivan Industrial IoT programme manager, offered insight into the potential impact of AI on the manufacturing landscape.

Defined as the science behind developing intelligent machines, AI is an advanced form of computer technology that empowers machines to perform tasks that are normally workable only by humans. Also known as cognitive intelligence, AI is the umbrella term used for a variety of underlying technologies including machine learning, deep learning computer vision, speech recognition, robotics, natural language processing, and deep learning. 

“Cognitive intelligence is not just about giving manufacturers the ability to gain answers to known questions; it is also about empowering industry to find new answers to emerging questions in a similar way to Similar to how earlier revolutions in manufacturing have seen several benefits from lean manufacturing, automation and IT, AI looks very promising as the next lynchpin for Industry 4.0,” explained Sundaram.

For manufacturing Sundaram expects deep learning to be the most important AI development as it is about enabling a machine to understand and programme itself using a complex set of algorithms. “This could be the resolution to a big issue currently facing manufacturing today to manage the huge amounts of data that are now being collected from production processes. Deep learning can be used for analysis of this data.”

Frost & Sullivan expects that deep learning, along with neural networks, will be the largest contributors to the creation of cognitive factories and we could well expect to see such factories by 2020. 

Frost and Sullivan is seeing a big shift, as Sundaram explained. “Traditionally, sciences have been based on a very simple logic construct ¬– a process resulting in an effect. However, today science is having to confront that fact that there is a complexity of multiple causes having multiple effects.  It is no longer possible to simply reduce one cause to a given  effect and this is where AI has a huge role to play, helping to resolve this complexity and giving science a whole new dimension.”

Sundaram went on to explain that AI could also help manufacturers to improve their processes. “Cognitive intelligence is vital to the future factory architecture – it is one of the central pivots around which the future factory will be built. When it comes to data analytics and predictive maintenance, AI really is the way to go. Siemens, for example, has recently added AI to its Simatic controllers which will allow them to program themselves.” 

But, what actually happens when a machine gains cognitive functions? Sundaram went on to offer some examples. “A machine with the benefit of computer vision is able to provide remote analysis of product designs and can visually recognise errors. Give a machine the ability to understand spoken commands, in the natural language of the user, and it will offer something far superior to what we have today.” 

Sundaram also expects AI to change the way that products are designed, produced and sold in the future. “In the manufacturing sector, because AI gives the ability to understand the combination of many causes leading to many effects it will change the very foundation of applied logic as we know it today,” he said. 

Key applications are expected to be found in quality inspection, navigation and movements of goods on the factory floor, just-in-time replenishment of materials in assembly lines, situation-based analysis, cognitive cobots, biometric face recognition and machine part tracking.

Offering a real life use case Sundaram explained how Google has used AI technologies to reduce the power costs of its data centres by 40% by improving energy efficiency. He explained that similar techniques could be used to streamline energy efficiency in factories and process plants without the need for changes in the short-term. “It is just a matter of collecting data and applying a certain algorithm to understand how to  operate the factory more efficiently. Of course, there are already ways of achieving this, but AI’s benefits are exponential,” he said. 

It is expected that progress in cognitive applications for the manufacturing sector will be fostered via partnerships between technology vendors and OT companies. Currently investment in AI is higher among discrete industries than in process industries and, as more emerging technologies come to the fore – such a Blockchain, segmented reality and AI – it is expected that there will be a convergence of these trends to facilitate unmanned operations and fully-automated factories. “For cognitive factories to become a reality cross-industry collaboration, and technology partnerships along with mergers and acquisitions will be required,” said Sundaram.

Mainstream AI applications are expected by 2021 and a fully-fledged cognitive, lights-out facility is expected by 2023.  Sundaram predicts that China will be first to achieve this as it does not have the legacy plant infrastructure and it has the political capability to make it happen.

In conclusion, Sundaram explained that the current barrier to wider adoption of cognitive technologies in manufacturing is mostly down to the inconsistent data that is currently available. “Unless you harmonised data it is not possible to implement deep learning strategies,” he said. 

“There appears to be no limit to what manufacturing will be able to achieve with AI. The technology will be increasingly used for production, quality control, design time and material waste reduction, and predictive maintenance performance. Factories of the future will continue to learn, develop, and perform better.  We have now arrived at a point of time where it is not difficult to envision factories as utopian hives of automation. As cognitive technologies mature and costs drop, manufacturers will start discovering new applications of AI that will help them make complex business decisions. Even though there will be some displacement of jobs at the bottom level of automation, businesses will begin to focus on re-training these workers to perform higher levels of design, programming, or maintenance tasks.”

Contact Details and Archive...

Print this page | E-mail this page