AI in manufacturing - get inspired and find support!

Artificial intelligence (AI) is everywhere today. Although the term was first used as early as the 1950s, AI has gained increasing attention in recent years. The importance and potential impact of AI is estimated to be very high, also for the manufacturing industry. However, despite the great expectations, many manufacturing companies are struggling to match AI with their concrete challenges: should we take the leap, or should we wait? Agoria and Sirris will gladly guide you through this process!

What’s in a name?

Although the term Artificial Intelligence (AI) is known to most (and is often used), let us take a moment to go back to the basic definition. The term was first defined in the 1950s by Minsky and McCarthy. They described AI as: “Any task performed by a machine that would have previously been considered to require human intelligence.” Since then, this definition has been regularly amended or modified to better reflect progress in this area. Nevertheless, the basic definition still stands.

Looking at the common ground within the many descriptions of AI, some basic pillars can be found:

  • AI systems exhibit at least some of the following characteristics associated with human intelligence: reasoning, learning, problem solving and planning (and sometimes even creativity).
  • AI enables technical systems to perceive their environment, deal with what they perceive, solve problems and act to achieve a specific goal.
  • AI systems are able to adapt their behaviour to a certain extent, by analysing the effects of previous actions and working autonomously.

To put it in a simple way: “Artificial intelligence refers to the ability of a computer or machine to mimic the capabilities of the human brain (observing, recognising, learning...) and to combine these capabilities to perform functions that a human could perform.” (Source: IBM)

Potential for the manufacturing industry?

A recent study by BDVA, euRobotics, ELLIS, CLAIRE and EurAI shows that the manufacturing industry is in the top 5 application areas for AI (see table). Moreover, the manufacturing industry is also in the top 5 in terms of AI spending.

Top Sectors for AI, Robotics and Data (Source: Strategic Research, Innovation and Deployment Agenda AI, Data and Robotics Partnership Third release September 2020)

Overview of promising applications

So, there is definitely a far-reaching interest in AI and manufacturing companies are investing in AI technologies. But which are the most promising applications? Sirris and Agoria carried out an extensive study to map all this out. The results will be published after the summer. However, we will offer you a little sneak peek.

Based on analysis of the literature and inspiring examples, a number of application domains keep cropping up for the manufacturing industry. We will briefly explain some of them:

  • Predictive maintenance: AI-assisted predictive maintenance (PdM4.0) involves using AI to gain insights and detect patterns and anomalies that cannot be detected (or are more difficult to detect) by (expert) operators. The aim is to anticipate potential failures and errors. Machine-learning techniques are applied to e.g. 1) analyse the collected data in real-time, 2) find correlations between historical data and current measurements, 3) identify potential disturbances, 4) suggest risk-reducing actions and/or optimal maintenance strategies.
  • Generative-design software is used to generate design proposals based on a number of design requirements and parameters (e.g., mass, stiffness limitations, material and manufacturing methods). By using AI algorithms, the designer can reduce the ever-increasing complexity of the design phase: the software automates various tasks (data analysis, simulations, proposal generation), while the expert can focus on design aspects.
  • Demand planning is part of supply-chain management, in which the demand for products is predicted in order to subsequently organise production so that this demand can be met (e.g., production strategy, inventory strategy...). Demand planning involves a lot of calculation, data analysis and is repeated cycle after cycle. AI algorithms offer a clear advantage in automating and improving this process.
  • Machine vision/quality control: many processes require a (visual) quality control. Because human operators are limited in their (visual) ability to identify errors, so-called machine vision systems are often used (these are not only many times more sensitive than the human eyes but also faster because of their computing power). In traditional machine vision systems, potential errors are identified, but the expert still plays a crucial role in whether or not to reject them. AI-based systems go one step further and can 1) learn which aspects are important for quality (e.g. recognise defects), 2) create rules that determine the final quality and 3) make decisions autonomously (good/bad).

Participate in our group session!

Do you want to know more about the topic of artificial intelligence and the possibilities for the manufacturing industry? Sirris and Agoria are organising thematic sessions about this topic. The first session starts on 28 June. More information is available here.