Artificial Intelligence and the Fourth Industrial Revolution
We spoke with Nigel Moulton, Global CTO, Modern Data Center Business Unit, at Dell Technologies about AI’s place in 4IR:
Artificial Intelligence is essential to 4IR as a systemic element present in so much of its expressions. For example, it helps facilitate the design process with ‘generative design’ software to identify design opportunities as well as problems and supply a range of solutions during both the initial design and testing. Predictive maintenance uses algorithms to predict potential failures in specific elements of manufacturing and generating alerts. Similarly, quality control uses algorithms to monitor quality and identify quality defects in real time without suspending manufacture and generating relevant data for analysis. The safety and optimisation of robotics and human collaboration is also an area where AI has a place. Collaboration is, indeed, a major element of manufacturing to which AI contributes, enabling customized solutions as the result of the interaction of different specialists and specialisms. AI extends beyond the factory to optimise the supply chain and distribution network as well.
Nigel Moulton, Global CTO, Modern Data Center Business Unit, Dell Tehnologies, believes that AI will play a pivotal role in 4IR as the technology will be deployed to provide insight that will enable manufacturers to adapt industrial and business processes to become more efficient, or safer, or both: “There is no ‘one size fits all’ approach as each company will use a combination of AI and IoT generated sensor data to shed light and better understand a process. The intended outcome will however determine the way technology is employed, whether that’s an AI to sense, reason (infer) and adapt to produce an output that drives real-time reaction; augmented by Machine Learning (ML) to store and analyse the AI data to improve the outcome over time; or Deep Learning (DL) to infer outcomes through multiple algorithms and data sets.”
When and how much
Can AI be applied piecemeal? Can a manufacturing plant become partly smart, but remain partly pre-4IR as companies implement new technologies according to their budgets, idiosyncrasies and the limits of their imagination? The answer probably lies in bespoke solutions that, nevertheless, embrace the need to build a smooth integrated manufacturing framework that works and will continue to work for years to come. This would imply the application of AI with the intention of evolving eventually into a complete system where the revolution has no more corners into which it has not gone. Each company is different and has its own priorities, which are reflected in its practices, policies and infrastructure. It is worth, therefore, investing time as much as money into determining the best course and most appropriate time schedule for complete adoption.
“There is an emerging use case for applying smart technologies to a discrete process or supply chain to help quantify, codify and then scale best practice from a ‘known good’ throughout the organisation,” reveals Moulton. “In this case, smart technologies can instrument and measure something that you already do really well, help to quantify why and how it’s a core competency and then to map that competency throughout the entire supply chain process. This may well include bringing in third party suppliers, encouraging them to be a part of the instrumentation process.”
Moulton notes that businesses that leverage the intimate knowledge of the manufacturing processes held by factory-level workers to identify feasible AI initiatives to make targeted improvements to productivity tend to experience better results. “This approach, combined with partnering with trusted technology vendors, can significantly shorten the path to becoming an AI-driven company,” he says.
Benefits of AI
AI is about partnership and integration. The 4th Industrial Revolution is digitally driven and shows no signs of slowing down. The potential for AI application is far from exhausted but continues to supply answers to new problems as they arise and new areas of manufacturing as they emerge.
Data and connectivity are of prime importance. Data whether live or historic, much gathered by sensors and processed remotely, needs collection, collation and analysis. AI makes manufacturing more flexible by introducing prediction and reactivity, responding to eventualities and providing solutions.
The 4th Industrial Revolution involves a fresh layering of manufacturing input – human, mechanical, electrical, digital and robotic. AI is the ubiquitous agent of change and facilitation whereby the establishment and integration of the layers are established to comprise a single manufacturing entity. Whereas mundane labour tasks will be performed increasingly by machines rather than human beings, the human workforce will be redeployed rather than replaced. The smart factory has become a machine itself, run by AI and requiring: minding, maintaining, administering and monitoring by human beings, but ever more remotely.
Generative design and training
AI-driven generative design is an evolution of AI-human partnership where we set design objectives for the AI to consider, explains Moulton. “This approach can significantly accelerate the development time and reduce the overall cost of manufacturing dramatically. But it’s worth noting that the human ultimately remains in charge making the final design decision.”
Monitoring and testing in real time?
“AI systems deployed in conjunction with IoT sensors can repeatedly measure and react with extraordinary speed and accuracy across a huge spectrum of environments,” the Global CTO comments. “But AI is only as good as the algorithms that it is deployed with. Perhaps surprisingly, much depends on the definition of real time, the associated thresholds, how they are measured and what constitutes a deviation. When combined with techniques such as Machine Learning AI systems can in some instances infer outcomes that are too difficult or nuanced to be detected by human monitoring techniques.”
AI in the UK
The UK is a world leader in AI and 4IR, so well placed as a global smart manufacturing hub. Innovative companies are already located at the beating heart of technological innovation and primed to be at the forefront of change. The UK government has set out its vision and strategy around Artificial Intelligence as a ‘grand challenge’, says Moulton. “This is why AI is now taking centre-stage for the UK government’s industrial strategy focusing on clean growth and the future of mobility. The strategy also includes opportunities to look after and managing an aging society and how to exploit the datasets in government and the private sector using AI to improve productivity and generate economic value. The UK is in great shape to explore these challenges because we have world-leading universities and leading private businesses. We also have some of the most digitally minded consumers and businesses in the world that has firmly placed our country as one of the leading nations around digital transformation.”
“The next phase of AI in manufacturing is reaching scale to deliver the economics of mass instrumentation of an entire manufacturing environment,” says Moulton. “AI as a technology has matured to the point where it should not be considered a science project, and the costs associated with the required instrumentation engines – IoT connected sensors – are at a point where they can be considered marginal versus the returns that can be realised. As companies start to recognise the process improvements and the associated returns to the bottom line of a well instrumented, constantly iterating industrial process AI, ML and DL will become embedded in business practices and viewed as commonplace. In fact, it will be considered retrograde to not have them.”
Product Designs Beyond the Human Imagination
Why future product development with generative design and artificial intelligence will now gain momentum
Author: Paul Haimes, VP, Europe Technical Sales at PTC
The concept of generative design controlled by artificial intelligence (AI) in product development is not ground-breaking innovation. Some years ago, there was the first hype around this topic; but there was still no real breakthrough in this field. The scepticism towards AI algorithms seemed to be too great a driving force in product design. In addition, it was difficult for suppliers of 3D CAD technology to get engineers and developers enthusiastic about the early and frequent use of simulation technology in the design process, as the technological challenges such as the lack of application speed often seemed too great. But, a lot has happened since that time. Thanks to cloud technologies and advances in simulation technology, 3D printing, and, as one of the AI technologies, machine learning, the concept of generative design in product development is facing a broader application in the market. And with that, it is now time to think about the future role of the product developer.
The true “Computer Aided Design
Simply put, generative design means humans and computers working together to create objects beyond the human imagination alone. While product developers and engineers have been actively thinking and creating a new product, component or larger construction – such as a bridge or a house – using the computer and modern CAD software as a tool, generative design turns the computer into the driving creative force. Human beings first define design parameters and functional requirements such as maximum size (installation space), weight, type of material, load capacity, manufacturing process or costs. It is even possible to define further design parameters that consider purchasing decisions, manufacturing capacities, the status of the supply chain and regionally required product variants. Then the computer takes over, but it not only calculates an optimal geometry, it also creates thousands of design drafts that meet the specified criteria catalogue, pushing it in all possible directions. Thus, simulation is integrated into the development process through generative design. So, for example, the system only generates designs suitable for the CNC milling machine or the 3D printer, although differing production methods can be proposed for selection.
The advantages that companies achieve through the generative design process alone are manifold: The productivity of the product development department in the design phase increases immensely as well as the gain in creativity and innovation if more time can be invested in researching conceptual designs. For example, it is possible to develop more powerful designs with lower weight and improved durability. At the same time, this type of development promotes the optimization of new products for improved manufacturability, the reduction of material costs and shorter production times, and it allows a high degree of personalization, which will delight customers. Because simulation, analysis and manufacturing are all on the same level, the risk of costly rework is dramatically reduced, which can further reduce time to market.
AI chooses the “winner”
After several thousand design drafts have been developed, using generative design and a possible catalogue of boundary conditions, technology does not leave the human being alone – how should he finally determine which draft fits best? This part of the evaluation is carried out by AI. It selects the most suitable version and thus selects the “winner” based on the specifications. The developer, on the other hand, can have several design variants selected at this point using various parameters such as the best suggestions for different materials ie the lightest model with the greatest possible stiffness. Criteria can also be changed in real time, whether it’s material or design requirements, as well as parameters linked to production costs, such as production volume, are converted immediately by the software.
The technology provides the developer with optimized designs for several targets simultaneously in a very short time. Once the possibilities have been explored, the first prototypes can be produced in milling machines or 3D printers, or the results can be automatically incorporated into tests based on other company findings, including cost calculations, supply chains and quality data. This not only saves an enormous amount of time, but also greatly increases output. The “aided” in CAD – Computer Aided Design – is finally actually implemented.D
Generative design is not topology optimization
At this point, it is important to review the difference between generative design and other technologies such as topology optimization, grid optimization or the like, which are often listed under one name. While generative design is based on a “white sheet of paper” for which the product developer only defines a few framework criteria, all other technologies refer to the optimization of an already existing design through simulation on the 3D model. Here, for example, the aim is to reduce the weight without changing the outer shape of the component. However, this does not create completely new design possibilities as in generative design, but only optimized variations of a known solution.
The product developer remains the last instance
As already described, generative design and AI change the entire product development process and thus also the role of the product developer or engineer himself. In the past, he was the driving creative force that delivered design drafts but in this process, he becomes more of a curator of the results. Although he remains involved in the design phase from the outset, he only defines the parameters for the computer and then juggles with a few objectives such as the fastest production variant, the most cost-effective model in production or the variant with the best product characteristics. Compared to today’s reality, this has the advantage, among other things, that he no longer must defend his models against other decision-makers; after all, all of them have been created using algorithms, whereby the respective optimum can be assumed.
However, an important role remains – the examining eye in terms of optics and aesthetics. The computer does not (yet) possess this. It calculates all given parameters and implements the optimal design according to technical aspects. Does a car ultimately have an appearance that appeals to the buyer despite its “optimal” design, or does it become a shelf warmer for car dealerships? What about colour and people’s aesthetic sense of form? Does a material feel “better” in the calculation despite its poor performance and is it more likely to be bought? These are all questions that demand the product developer’s broad set of multisensory capabilities and experience. There is a difference whether it is a component within a machine, car or airplane that no one will see later, or whether it is a component or a product that is visible later and whose shape, colour or sound can be decisive for the purchase.
Ready for daily use
Generative design and AI have now reached a maturity that makes a wide range of applications possible. This is also ensured by new application packages, such as PTC who offer it within its future Creo portfolio, in which the ANSYS visualization technology and the Generative Design technology of the company acquired at the end of last year are frustrating. The joint solution shifts the analysis to the beginning of the design process. Using the integrated functions of Frustum and ANSYS, Creo can recommend design methods with generative design, guide users through the iterative design process with ANSYS Discovery Live and validate the complete product to scale with the more comprehensive ANSYS Discovery package. These Creo-integrated features provide product developers with world-class opportunities to rapidly drive product innovation.
Gone are the days of slow applications that cost more time than they save. As a result, generative design and AI are likely to be more widely adopted very soon as companies increasingly discover the benefits of this method in conjunction with available technology. Especially the first users who will be able to design and manufacture products faster and offer them with improved features or at a lower price.