The smart factory is becoming a reality, as manufacturers take advantage of the latest in machine learning, big data, and robotics to achieve new levels of efficiency and competitiveness. Michael Nelson explores the ways in which artificial intelligence (AI) and automation are playing a key role in this transition.
The manufacturing landscape is currently undergoing a metamorphosis in its drive to increase production efficiency and reduce cost, thanks in no small part to the advent of AI. This rapid advancement in technology is helping to usher in a Fourth Industrial Revolution, based on large-scale machine-to-machine communication and the internet of things (IoT).
According to a McKinsey report into manufacturing supply chains, AI is improving demand forecasting by reducing forecasting errors by 50 per cent and reducing lost sales by 65 per cent with better product availability. In other aspects of manufacturing too, AI is transforming performance capabilities.
“If you look at what you have to do throughout the plant lifecycle, people need to understand much more about the work environment and how applications should work,” says Bernhard Eschermann, CTO of process automation at ABB’s industrial automation division. “This is where AI comes in. In a typical use case in operations, you would usually have people sitting in a control room which runs for 24 hours a day, seven days a week. They will be trained to observe processes happening on their screens and identify when certain limits are exceeded or when alarms are sounded. But if that situation were to change so that the control room was not in operation full-time, or the people observing operations were less experienced in that environment due to managing multiple sites, the in-depth knowledge and context of that plant will be lost.
“AI can help in this case, processing information and identifying when there are irregularities in production before there is an alarm. It can also analyse reasons for why the production process went wrong, and what can be done to fix it.”
While it may be tempting to invest in a fully instrumented, modern plant with multiple AI applications, Eschermann says that in most cases this is not something that is particularly necessary in order to maximise return on investment (ROI). Advances in AI technology mean that it can be adopted much easier and in incremental stages, meaning that it is becoming much more accessible for manufacturers with limited resources.
In process industries, for example, where it does not make sense to change the basic set-up of pipes and vessels, non-invasive sensors, which can measure the temperature of a pipe from the outside, are replacing traditional temperature sensors. Previously, these traditional sensors were inserted in to the pipe via a hole which was drilled in to the casing, meaning that production would have to stop during installation. This increased the amount of downtime experienced by manufacturers, as well as the cost. Non-invasive sensors mean that production continues during their installation, and costs attributed to downtime are reduced.
“To make AI cost effective, businesses should first identify what the problem is that they want to solve, and then figure out what they can do to solve it,” says Eschermann.
“This depends partly on how difficult it is to get access to the data you need, but also on setting a clear business case for its implementation – understanding how to manage the constraints of the business, leading to a manageable investment. If you can then scale that, that can even help to increase the return without increasing the investment by very much.”
AI advances achieved as a result of pandemic-induced innovation
In major global companies, decisions by senior executives have typically been based on past events within that company, along with manual forecasting on what might happen next. However, the pandemic has highlighted the need to build and anticipate risks across the whole supply chain.
“Organisations must now take steps to embed AI and data science to make predictions for the future – and based on more data than they ever had before,” says Rafi Billurcu, partner and head of manufacturing at Infosys Consulting. “With datasets going back years – and the need to look at external data like weather records, demographic changes, or government policies – it is only possible to really leverage all this data with the helping hand of machine learning and AI.”
Machine learning (ML) algorithms learn from themselves over time, so they become more and more accurate with each new piece of data that is added. This gives executives a completely new, real-time view of their business, enabling them to see correlations and make predictions that were impossible to see before, helping them to propel their business forward.
“AI and ML technology is not exclusive to big manufacturers either,” continues Billurcu. “For small manufacturers – often ones with more constraints on resources – it is especially important to be smart and data-driven, and to increase visibility to identify opportunities for improvement. Even more basic automation technologies can suffice, as long as they are implemented in a more agile way which allows end-to-end visibility across the overall performance.”
One of the ways in which manufacturers have incorporated AI is through moving factory operations to a ‘hands-free’ model, futureproofing themselves through innovation. Sectors like supply chain, logistics, and manufacturing have long been known for relying on manual, paper-based processes – and for lagging behind their more forward-thinking counterparts. The pandemic changed all that.
“Remote operations have been critical to minimising disruption during the pandemic,” explains Billurcu. “But the benefits go far beyond accommodating social distancing. The application of automation and other Industry 4.0 technologies such as AR, VR and analytics ensures efficiencies can be achieved with minimal impact, strengthening both organisations’ top and bottom lines.”
‘Hands-free’ operations enable senior teams to work in predictive mode – to predict future trends and events within their supply chain – and even prescriptive mode, where AI can recommend solutions and advise on the best course of action. Here, AI and automation can enable modern factories to weather potential storms.
Moving forward, Billurcu concludes that it will be essential for manufacturers to continue to inject AI and automation into operations, making them as hands-free as possible. “Doing this will ensure business continuity no matter what, and free up more workers to focus on higher-level tasks. In this, human interventions should be minimal and only on a need-to-know basis.”
Automation as an alternate solution to AI in driving efficiency
While the ability to use AI to create efficiencies which were previously unachievable is something which industry is currently striving towards, simple automation may be just as effective. Mark Gray is the country manager at industrial robotics manufacturer Universal Robots. He says that some manufacturing tasks can be simplified without the need for any kind of artificial intelligence.
“As a concept, AI uses a lot of data to learn about different manufacturing processes. However, those processes can be boiled down to really simple, specific tasks which do not need to use AI. For example, if you were to walk around a factory, you would see people carrying out tasks using their hands and arms, whether that be transferring parts in to and out of machinery, carrying things, or any other manual labour. Automating these types of manufacturing jobs can add value to a manufacturing business, particularly small and medium sized businesses (SMEs), without the need for all the machine learning algorithms and AI which would obviously add to cost.
“Our UR5 collaborative robot (cobot) arm shares the same sort of dimensions as a human arm and can be utilised in an automated system to share the workload in collaboration with human workers.”
Collaboration is a key component in modern manufacturing, where humans and cobots work together to increase efficiency and produce better quality products.
“We talk about empowerment through robotics, and when people and robots work together, you get the best of both worlds,” continues Gray. “Robots can take heavy or repetitive tasks away from people who are in danger of becoming injured while working. An added benefit of this is that it decreases idle time, as robots work on repeatable processes alongside their human counterparts in a shared workspace. In this way, robots are used more like a tool.
“On the flip side, human workers are free to bring creativity and intuition to the work cell they share with cobots. They can focus their time and attention on taking care over the products they make, increasing their quality, as well as use their experience to improve manufacturing processes. By balancing tasks, humans can add more value to their work cell.”
Ian Ferguson, vice president of marketing and strategic alliances at Lynx Software Technologies, says that while cobots are currently a very small percentage of the overall robotics market, it is an area which will grow rapidly in the next five years.
“Cobots need much closer control for real-time implementation of complex decisions in co-working environments. This is an area where there is a lot of focus on AI to improve the user experience with these types of machines. This decision making must be made by the robot as an edge device in order to achieve the speed and latency to cope with increasing data from more IoT sensors, and the consequences of getting a decision wrong. The more pioneering manufacturing plants, however, are starting to rethink processes to make more efficient use of humans and robots together.”
Future advances in robotics
Innovation in robotics shows no signs of slowing down. One way in which robots are expected to advance moving forward is in the integration of sensors into industrial robots.
“Human beings have five senses,” explains Ferguson. “Today, robots mostly just use one: vision. The capabilities of these systems will continue to improve, and I can envision machines beginning to understand gestures, for example, in settings where there is a lot of noise. But I expect to see robots adding the other senses.
“Future robots will be able to listen to voice commands and identify noises that indicate something in the manufacturing plant is not behaving in a usual way. They may also be able to use touch to confirm a particular product is smooth enough, and perhaps they could taste test the mixtures of specific compounds.
“The only sense that may be more limited in robotics is smell, but some companies are working on sensors for this function, with fascinating future applications. For example, the smell of urine has been found to be a leading indicator for certain cancers, and it has been proven that bees can smell explosives. Imagine an agricultural setting where fruit can be selected based on their ripeness.”
While automated robotics can augment human labour in manufacturing and replace humans in jobs which are repetitive or dangerous, the potentially complex robotic procedures of the future envisioned by Ferguson, suggests that machine learning and AI will continue to be at the forefront of innovation. It is up to individual manufacturers to consider the cost-effectiveness of implementing this technology when compared to their business objectives.