BMW Group plans to open a new electric vehicle plant in Debrecen, Hungary, in 2025. By the time the factory goes online, the facility’s layout, robotics, logistics systems and other key functions will already have been finely tuned, thanks to real-time simulations using digital twins.
It’s the world’s first “digital-first” factory and a striking example of the ongoing and growing strategic pursuit of digitalization by manufacturers worldwide. AI is a key part of many efforts. Advances in intelligent technologies and products are enabling new or improved use cases across the manufacturing lifecycle, from product design to engineering to fabrication, testing and assembly.
Smartly adapting to changing market and consumer needs
The rise of \”Industrial AI\” is being led by digital-first factories. Industry 4.0 was gaining traction as a way to speed up and restructure production before COVID-19. The strategy\’s goal is to make the most of cutting-edge innovations in areas like as data science, artificial intelligence, cloud computing, robots, the IIoT, HMI, clean energy, cutting-edge engineering, and more.
Today\’s manufacturers continue to invest in intelligent technology and infrastructure as important foundations of \”smart manufacturing\” in response to economic uncertainties and chronic supply and labour constraints.
According to IDC, worldwide sales of artificial intelligence will total $154 billion in 2023, with manufacturer investments accounting for a whopping 16.6 percent of the total.
Naturally, each business has its own unique AI objectives. To be more efficient now and more competitive in the future, manufacturers are increasingly using smart technology. And, of course, to meet the ever-evolving demands of the market and your customers. Most people are looking for help in three main areas:
- Smarter manufacturing aids in improving accuracy, output, and quality while decreasing costs.
- Faster product development, enhanced performance analysis, and a more adaptable, dependable supply chain are all possible thanks to increased agility.
- A more sustainable society means lower energy bills and less of an influence on the environment.
The latter is becoming more crucial. Environmental, social, and governance (ESG) standards are becoming more difficult and time-consuming for many businesses to meet. Smart factories and manufacturers may decrease consumption, pollution, and waste while enhancing material recycling by utilising less resources. Logistics and shipping routes may be improved with the aid of AI. The development of more eco-friendly materials may be streamlined with the use of generative methods.
Cases of new and improved utility
How do companies intend to reap these advantages? Investing heavily in maintenance and quality analytics is common in both live and planned systems. In-progress implementations include digital factory twins.
Predictive maintenance powered by AI is an alternative to traditional scheduled or time-based methods for avoiding breakdowns. Here, GPU-accelerated processing is used to analyse massive volumes of sensor and operational data in real time, allowing manufacturers to foresee breakdowns and plan repairs with increased precision and speed. False positives and negatives may be drastically cut down on with AI-driven preventative maintenance. Engineers may use the data to identify possible problems and implement fixes to avoid similar situations in the future.
READ ALSO Next-gen driving: Smarter, software-defined cars that can be upgraded like phones
Many businesses see QA/QI as a key AI priority. The American Society for Quality (ASQ) reports that this is because faults account for roughly 20% of manufacturers\’ total sales income. Product recalls and warranty costs rise when quality is compromised, and so does the long-term harm to a brand\’s reputation.
Many factories now use computer vision programmes powered by artificial intelligence to aid with fault detection. However, current automated optical inspection (AOI) equipment need for substantial investment and human engagement. The quality of produced parts should be enhanced by new approaches that promise to make better use of AI and ML. Cracks, chipped paint, improper installation, loose joints, and even foreign substances like dust and hair are no match for their keen eyes.
An innovative method currently under development leverages object perception and synthetic data to rapidly train fault detection algorithms.
The effectiveness and robustness of the supply chain
As the COVID-19 epidemic tragically shown, many businesses are ill-equipped to deal with sudden shifts in production and distribution. To this day, there are still global shortages of everything from toilet paper to semiconductors. Seventy-two percent of firms surveyed said that supply chain disruptions and shortages of components are their major source of worry for 2023. The prolonged lead times of shipments continue to be a major cause for worry.
As a result, over 90% of supply chain experts aim to invest in making their supply chains more robust, notably via the use of cloud. For more accurate demand and supply forecasting, more efficient logistics and transportation route optimisation, and smoother supplier and distributor coordination, many firms are turning to data analytics and AI/ML. The purpose is to enhance productivity and responsiveness in order to forestall or mitigate interruptions.
The most prepared businesses will use artificial intelligence (AI) and safe, scalable cloud computing to thrive in the new normal. The ability to execute in the cloud or on the edge, together with increased service levels and lower costs, is only some of the benefits of better planning and optimisation. If manufacturers had better end-to-end insight, they could leverage supply and demand signals to reduce uncertainty and maximise possibilities.
Discouraged by Intricacy and Distant Information
While the manufacturing industry is increasing its spending on digital and data infrastructure, it is still behind other sectors in its adoption of operational AI.
Only 10% of 700 organisations across the globe examined by PwC had finished or were in the late stages of their digital factory implementations, likely due to challenges putting AI into production at scale. Almost two-thirds were either unable to demonstrate any results at all or were just getting started on their digital journey.
Researchers have identified very diversified and scattered machine landscapes and complicated system settings as primary causes. Many businesses have difficulty expanding the use of isolated solutions throughout their whole production infrastructure. Progress is sometimes stymied by the high expense of implementation as well. Many of these issues stem from the requirement to integrate and optimise a specialised technological stack consisting of hardware, software, skills, and infrastructure.
There\’s also information. Both discrete and process manufacturers have put significant resources into laying the digital groundwork for a smart factory during the last two decades. Data from machine control systems, videos/surveillance, the Internet of Things, and other sources was collected by new technologies and instruments, and is now flooding into analytic and AI platforms.
However, more information is not necessarily better. Many executives in the business and technology sectors still have difficulty extracting useful information from the vast oceans of operational and information technology data. Problems with data quality, accessibility, and centralization are pervasive, adding to the difficulty.
Progress is possible thanks to technological development
However, manufacturers may perhaps get over all of these obstacles with the aid of cutting-edge cloud technologies that have already been demonstrated in the field.
READ ALSO Collaboration between Relational AI and Snowflake will transform business AI decision-making
Traditional IT infrastructure (including processing, storage, networks, development environment, frameworks, software, and virtualization) cannot keep up with the exploding data volumes, complicated parallelism requirements, and other characteristics of industrial AI workloads.
\”AI-first\” systems and toolkits are tailored specifically for artificial intelligence. As a result, manufacturers have access to pre-integrated platforms and models that can streamline and quicken deployment from the edge to the cloud, all while freeing up valuable resources to be put towards more meaningful data science. Integrating disparate data sets is much simplified in an end-to-end, full-stack setting. It offers a framework for making data consumable and actionable in real time, facilitating both decision making and model training at every stage of the AI manufacturing pipeline. Consulting firm PWC agrees that a standardised digital backbone is a crucial element in modernising factories.
Cloud-based AI infrastructure is essential for many factories to implement smart, industrial processes at scale. With this method, businesses may save money and gain new skills without having to shell out a tonne of money for expensive infrastructure upgrades. Accenture claims that compared to on-premise deployment on underutilised systems, the savings from moving or creating AI infrastructure with flexible, pay-as-you-go cloud services may be as high as 20-40%. Those cost reductions don\’t account for potential gains from less energy use and more efficient use of available space. Accenture also claims that manufacturers\’ operational risk is reduced when quality assurance and training can be moved easily to non-production locations.
Many attempts at artificial intelligence (AI) grind to a halt due to insufficient computational performance. Training takes longer due to slow processing, delaying time-to-value. Real-time urgency and sophisticated LLMs only make matters worse. The use of high-performance computing (HPC) may cut training time for artificial intelligence by as much as 20 times throughout all stages of production.
Supercomputing is becoming more accessible to manufacturers because to cloud computing. To train models for generative AI and other data-intensive applications, it gives instant, flexible access to the necessary supercomputing equipment and tools.
Microsoft and NVIDIA have released a product that makes supercomputing available whenever it\’s needed. Access to the infrastructure, software, and processing capacity required to train, construct, and deploy sophisticated AI models and applications from cloud to edge is immediately accessible to businesses across the world and is billed on a monthly basis.
READ ALSO Salesforce releases Sales GPT and Service GPT to streamline client interactions using generative AI
Smart manufacturing applications and \”digital-first\” facilities, like BMW\’s, need a link between the real and virtual worlds of production. The growing \”Industrial Metaverse\” allows for the real-time automation, simulation, adjustment, and prediction of AI-driven business processes by connecting data from physical sensors to their digital counterparts. According to Deloitte, one in five manufacturers are already testing or creating a metaverse platform or solution for their own goods.
Companies may now more easily use the metaverse for smart manufacturing thanks to new services. Developers now have instantaneous entry to a full-stack, native, and agnostic environment thanks to the platform-as-a-service (PaaS) NVIDIA Omniverse Cloud. Manufacturers may create and run precise, dynamic, fully-functional 3D digital twins by connecting with Azure Digital Twins and Internet of Things cloud services. The cloud infrastructure and capabilities, like as security, identity, and storage, required to deliver these corporate services at scale are all made available by Azure, just as they are for supercomputing services.
The enhanced capacity to digitally monitor, simulate, control, and operate physical assets is a boon to manufacturers. This allows for earlier problem detection and quicker course corrections, as well as more transparency into operational performance.
READ ALSO Online vs. In-Person: Where Should You Get Your Auto Insurance Quote?
Remote teams can now work together in real time using video, audio, and simulations thanks to the integration of 3D platforms with Microsoft 365 Teams, OneDrive, and SharePoint. Accenture recently showcased an excellent early attempt to reduce the amount of time it takes to make a choice, put that decision into action, and get feedback.
Technicians at service centres, for example, may one day use AR glasses to do complex repairs in a virtual environment, collaborating with other specialists through digital twins as the technology develops.
A new technology developed by a German business can take 3D data and turn it into scalable apps and interactive experiences for manufacturers. Developers may use Instant3DHub to create, launch, execute, and automate apps with \”any data, any device, any size.\”
And it\’s becoming clear that generative AI may be used to improve industrial automation and operations in areas like software engineering, defect reporting, and quality control inspection. Siemens and Microsoft have developed a proof of concept that demonstrates how plant employees and others may utilise mobile devices and natural speech to record and report problems with production, quality, or design.
Smarter is, simply put, better
It\’s not necessary for every company to be on the cutting edge of artificial intelligence. Artificial intelligence and computer simulation, however, are of immense use to everyone. Manufacturers and others might define \”smart\” as the achievement of higher quality, more efficiency, stronger supply chains, and faster time to value and innovation.
GPU-powered Azure cloud infrastructure and solutions from Microsoft and NVIDIA provide manufacturers with real-time acceleration, predictability, robustness, and sustainability.
READ ALSO Talent Select AI assesses interviewees for psychological and personality qualities