How Machine Learning is Transforming Manufacturing

Artificial IntelligenceInnovation
Industry 4.0

Manufacturing is transforming thanks to AI. As factories become more intelligent, their use of machine learning will only increase. Are we on the cusp of a new age of manufacturing?

Manufacturing is changing out of all recognition thanks to AI and specifically machine learning. Coupled with advanced digital vision systems, IoT (Internet of Things) and the imminent rollout of 5G, manufacturing is about to enjoy a perfect storm that will overshadow the impact 3D printing has already had across the industry.

Cognitive manufacturing is how IBM describes how this industry is now using advanced information-based systems. “Cognitive manufacturing is powerful because it combines sensor-based information with machine learning and other artificial intelligence capabilities to find patterns in structured and unstructured data from plant, enterprise and industry systems,” IBM state in their whitepaper.

“It pulls relevant information together in real-time and applies analytics to yield unprecedented levels of understanding and insights about the manufacturing process. It automates responses based on its findings and delivers actionable information as well as continuously updated knowledge to decision-makers in the manufacturing setting.”

Machine learning becomes the core around which new and advanced manufacturing processes can be built. Anupam Aishwarya, AVP solutions consulting at E2open explained to Silicon: “P&G uses AI to better sense and predict demand, using real-time data from its retail partners and factoring into its planning public data around holidays and events. Accurate demand predictions lead to more accurate production and distribution processes, enabling P&G to better serve changing customer preferences.”

Machine intelligence

Machine learning, robotic process automation and machine vision all have one thing in common: data. Information is at the core of all these systems. Machine learning, in particular, is based upon vast datasets, which can be used to deliver insights into a manufacturing process. Add a raft of new sensor technologies that will be coming online soon; manufacturers need to become masters of data collection and analysis.

“Cognitive manufacturing is all about exploiting data from diverse sources—not only equipment sensors but also logs, manuals, employee biometric monitors or the environment,” says IBM. “The approach incorporates these types of sources and data into the analytical process to create a knowledgeable system that is continuously learning. It can make insightful operational recommendations based on a comprehensive understanding of manufacturing conditions.”

Industry 4.0 has the potential to be a powerful driver of economic growth, predicted to add between $500 billion-$1.5 trillion in value to the global economy between 2018 and 2022, according to Capgemini.

Peter Pugh-Jones, Head of IoT Operations, EMEA & AP at SAS, commented to Silicon: “Data isn’t valuable – insight is. AI gives manufacturers the power to turn the huge amounts of information they now possess into actionable insight in real-time. Not only that it can also be used to predict potential disruptions or spikes in demand. For example, if a machine begins to function abnormally, an AI algorithm could not only flag it as something needing human attention but also offer predictions of how soon a fault might occur and what maintenance work might be required to rectify the problem.”

It is, though, critical that all manufacturers place the burgeoning machine learning services becoming available, within the context of their business’s strategy. Here, technology for its own sake won’t deliver the results expected, as Bob De Caux, VP AI and RPA, IFS explained: “Falling for the hype of AI is easy, but instead of taking a sudden leap into new AI territory, businesses must first lay the groundwork. CIOs and CTOs need to first gain a holistic, centralised view of their data assets and to think about what parameters to take into account, what are the factors that affect production, sales and forecasting in the past, are doing so now, and could potentially impact these things in the future.

“Once the necessary data has been captured, organisations can then look to ERP solutions which integrate this from various departments and sources, into a unified, centralised view. That solution, once designed, developed, and deployed must be able to gather, process, and analyse data in real-time from across every endpoint, device and sensor to produce actionable insights.”

Created by machine

The advances in machine learning will enable businesses to rapidly evolve their production processes. As Microsoft pointed out point out, robotics will be transformed: “The use of robots in manufacturing is nothing new, but this new generation of cobots is not your average machine. Today’s cobots are built with artificial intelligence and machine learning that power cognitive capabilities. These robots can use computer vision to quickly inspect large quantities of items for flaws, automate the transportation of materials throughout a facility and avoid hazards using predictive intelligence.”

Peter Pugh-Jones, Head of IoT Operations, EMEA & AP at SAS believes AI will herald a new age of predictive manufacturing: “The critical value that AI and machine learning bring to manufacturing is the ability to begin working predictively rather than reactively. Traditional business models are focused on responding to the past – what customers have been asking for, what competitors have been doing. By contrast, AI gives manufacturers the ability to act based on a deep level of analytical insight, pre-empting demand and challenges.

“AI can also fold in data from a wider variety of sources, picking up patterns across the back office, the shop floor and the supply chain, for example, that humans alone couldn’t spot,” Pugh-Jones continued. “Those patterns can, in turn, be used to inform strategy – whether that be changing the order list to accommodate an expected spike or changing production methods wholesale to respond to delays in the chain.”

Manufacturers have always been on the cusp of new technologies. Today, AI and its components – notably machine learning – will mean another quantum leap to make Industry 4.0 a reality. Manufacturers large and small will be able to use the data they have and, the masses of new information they can collect thanks to IoT, to transform their operations.

Silicon in Focus

Asheesh Mehra, Co-Founder and CEO, AntWorks.

Asheesh Mehra, Co-Founder and CEO, Antworks.
Asheesh Mehra, Co-Founder and CEO, Antworks.

Asheesh is the Co-founder and CEO of AntWorks and has over 20 years of experience working within the industry where he has established himself as a real heavyweight in the Robotic Process Automation (RPA) space, a component of AI and machine learning that has been disrupting the role of technology at enterprises across the world. AntWorks work with City Union Bank, Taxbot and Indecomm, so has some real-world expertise to bring to the mix.

How does AI form a significant component of Industry 4.0?

AI plays a significant role in Industry 4.0, as it elevates productivity, efficiency and progression in the manufacturing process. The smart factory which is made up of hyper-connected production processes comprising of various machines that all communicate with one another, rely on AI automation platforms to collect and analyse all types of data including images, standardised code text and categorised fixed field text, to optimise the manufacturing process. Essentially, AI is an essential component in smart factory operations, as it automates the whole manufacturing process, bringing that added efficiency and precision in all facets, from product design to production.

What areas of manufacturing are AI technologies impacting today?

AI is one of the leading emerging technologies already being utilised by manufacturers to improve product quality, drive efficiency and cut down operating costs, all of which are significantly improving the manufacturing process. One area that is benefiting from the use of AI in the manufacturing plants where we’re starting to see a working relationship between humans and robots.

Manufacturing companies that are looking to innovate and deploy AI need to look towards implementing intelligent automation platforms that use bots built with fractal science technology to undertake the more admin-based roles/tasks. These types of AI, known as multitenancy platforms, enable multiple bots to reside on a single machine and perform several different processes at once. It also allows personnel to use the same machine to perform other functions at the same time, meaning that less of the human touch is required to fulfil roles.

This, in turn, means that a significant number of workers can be reskilled and retrained at a faster rate, to take on the more technical and complex work such as designing and programming, effectively moving them off the production floor and on to higher positions. This is something that is already happening in some parts of the industry.

These bots will not only speed up the manufacturing process but also aid human workers in decision making. The bots can collect, process and analyse both structured and unstructured data in the form of algorithms and system messages in real-time.

Manufacturing warehouses mainly use unstructured data, like handwritten paperwork and inventory checklists, as part of their day-to-day workings; therefore intelligent automation platforms that are built on fractal science will be instrumental in transforming the modern warehouse. As a result of this, manufacturers will be able to cut down on production downtime while also optimising the overall operational efficiency of the manufacturing lines.

Who is using AI across their manufacturing processes?

Original Equipment Manufacturers (OEMs) that are successfully operating in smart factories and implementing industry 4.0 are using artificial intelligence in their production processes. Manufacturers who have undergone a digital transformation and can organise and utilise their data sets are taking advantage of the ability of AI and machine learning to improve quality control, standardisation, and maintenance through producing predictive analyses of equipment functionality and radically streamlining factory lines.

A majority of companies are now aiming to implement AI in their production processes, but far less have an AI development plan and to a greater extent, unsure of the appropriate type of automation platform to use.

How are CIOs and CTOs managing the implementation of AI within their manufacturing processes?

CIOs and CTOs are starting to buy into the smart factory revolution in their quest to achieve optimum operational efficiency in their manufacturing processes. However, a big part of their role in managing the deployment of AI is first to identify the processes that need automating. They must then decide on the right automation platform for achieving the set goals and provide a good ROI. This is a fundamental step in deploying and managing AI as the wrong decisions will undoubtedly lead to a waste of money, time and resources.

Aside from this, they also need to establish buy-in from workers at all levels and consider the effects the deployment process could have on their workloads, roles, system architecture and the current technologies at use on the manufacturing floor. This could also help pinpoint which roles need retraining so they can prepare accordingly.

What’s the future look like for AI across the manufacturing industry?

Manufacturing is an industry that is dependent on a range of regulations and laws from various jurisdictions. In the US, there are clearly defined Good Manufacturing Processes (GMP), which dictate the necessary measures that need to be applied to ensure systems and tech consistently produce quality and safe products.

In the UK, the General Product Safety Directive (GPSD) provides a strict safety framework to make sure that manufacturing processes and protocols adhere to very high standards. These strictures mean that any new form of technology that is implemented into a warehouse or manufacturing operation need to be audited appropriately and fit for purpose so that standards in the warehouse are not disregarded by installing new tech.

Regulating the AI application itself is the first step, but to prevent misuse of AI and automation platforms, regulations need to be integrated to avoid companies and governments taking advantage of new technology. AI has the potential to be used for malicious intent; from skewing election results to cyber-attacks, while it is mostly a force for good, there is also the potential for it to be used by malicious actors. This is why any AI used in a business must be ethical AI.

To avoid this misuse across the technology, the UK has already implied tech will face regulations to instil proper standards and use through rules that companies will have to abide by to instil quality, and guidelines have also been published with the introduction of the Centre of Data Ethics and Innovation and the Office for AI. With this movement already being made, we can expect regulations and rules on AI to be more widespread, and firms will likely develop technologies with regulations in mind.

In the near-term, we will see such regulation come to fruition, paving the way towards full automation adoption in the warehouse. It is these regulations that will not only set the pace for innovation in factories but will act as a barometer for the level of change workers are expected to see in the workplace.

Once regulatory bodies put the rules in place, manufacturing firms will begin to adopt automation on mass, and in this case, the most effective platform to install would be an intelligent automation platform built on fractal science. This form of automation will play a significant role in the digital transformation of manufacturing over the next decade or so. Warehouse processes such as inventory checks and shipment orders are often recorded through paperwork and other types of unstructured data. This data can be dealt with quickly and efficiently with integrated automation platforms (IAP) that use machine learning to consistently and coherently organise data like this.

With the true AI and Intelligent Automation (IA) market expected to grow from US$8 billion in 2019 to $14.4 billion by 2024, I think now is the time for manufacturing firms to consider how they move forward in this digital age.

Read also :
Click to read the authors bio  Click to hide the authors bio