Artificial Intelligence in Manufacturing

Industry 5.0 (I5.0) is a novel paradigm for the development and deployment of industrial applications based on Cyber-Physical Systems (CPS). It evolves Industry 4.0 in directions that exploit trustworthy human–AI interactions in human-in-the-loop scenarios. Despite the rising popularity of I5.0, there is still a lack of reference architectures (RAs) that outline the building blocks of I5.0 applications, along with the structuring principles for effectively integrating them in industrial systems. This chapter introduces a reference model for industrial applications that addresses critical elements and requirements of the I5.0, including human–robot collaboration, cybersecurity, safety, and trust. The model enhances state-of-the-art I4.0 Industrial Internet of Things (IIoT) architectures with human-centered I5.0 features and functionalities. Based on this model, the present chapter introduces a set of blueprints that could ease the development, deployment, and operation of I5.0 applications. These blueprints address technical integration, trustworthy operations, as well as the ever-important compliance to applicable regulations such as General Data Protection Regulation (GDPR) and the emerging AI Act.

Nowadays, the market for AI services is continuously growing and it is expected to exceed 5 trillion euros in the next 5 years. However, the sharing of knowledge is primarily achieved by the sharing of published AI-related papers. The sharing of the trained AI/ML models is still in its infancy stage and in some domains it does not even exist. In this chapter, a marketplace for exchanging AI models related to smart manufacturing and Industry 5.0 domains is introduced. The proposed AI Marketplace consists of a semantic-based repository that manages the AI models, a blockchain-based framework that adds the business logic and web-based user interfaces that enable models’ exploration and sharing, and transactions among the stakeholders. The purpose of this chapter is to present the implementation details of this AI Model Marketplace by highlighting the key concepts and technologies used along with the main supported functionalities. By using such a marketplace, the manufacturing companies are able to capitalize in a large variety of AI models to solve various problems enabling intelligent, flexible, and cost-effective production.

Modern manufacturing requires developing a framework of AI solutions that capture and process data from various sources including from human-AI collaboration. This chapter tries to describe the concept of domain knowledge fusion in human-AI collaboration for manufacturing. Human interaction with AI is enabled in such a way that the domain expert not only inspects the output of the AI model but also injects engineered knowledge in order to retrain AI models for iterative improvement. Domain knowledge fusion is a technique that involves combining knowledge from multiple domains or sources to produce a more complete solution by augmenting learned knowledge, i.e., the knowledge generated by the AI model with engineered knowledge, i.e., the knowledge provided by the domain expert. The concept developed in this chapter demonstrates how the domain expert interacts with AI systems to observe and decide the veracity of the learned knowledge with respect to the given context. It enables humans to collaborate with AI systems through intuitive interfaces that help domain experts in interpreting insights, validating the findings, and applying domain knowledge to gain a deeper understanding of the data.

The integration of Knowledge Graphs (KGs) in the manufacturing industry can significantly enhance the efficiency and flexibility of production lines and improve product quality. By integrating and contextualizing information about devices, equipment, production resources, location, usage, and related data, KGs can be a powerful operational tool. Moreover, KGs can contribute to the intelligence of manufacturing processes by providing insights into the complex and competitive manufacturing landscape. This research work presents a comprehensive analysis of the current trends utilizing KG in the manufacturing sector. We provide an overview of the state of the art in KG applications in manufacturing and highlight the critical issues that need to be addressed to enable a successful implementation. Our research aims to contribute to advancing KG technology in manufacturing and realizing its full potential to enhance manufacturing operations and competitiveness.

The representation and exploitation of semantics has been gaining popularity in recent research, as exemplified by the uptake of large language models in the field of Natural Language Processing (NLP) and knowledge graphs (KGs) in the Semantic Web. Although KGs are already employed in manufacturing to integrate and standardize domain knowledge, the generation and application of corresponding KG embeddings as lean feature representations of graph elements have yet to be extensively explored in this domain. Existing KGs in manufacturing often focus on top-level domain knowledge and thus ignore domain dynamics, or they lack interconnectedness, i.e., nodes primarily represent non-contextual data values with single adjacent edges, such as sensor measurements. Consequently, context-dependent KG embedding algorithms are either restricted to non-dynamic use cases or cannot be applied at all due to the given KG characteristics. Therefore, this work provides an overview of state-of-the-art KG embedding methods and their functionalities, identifying the lack of dynamic embedding formalisms and application scenarios as the key obstacles that hinder their implementation in manufacturing. Accordingly, we introduce an approach for dynamizing existing KG embeddings based on local embedding reconstructions. Furthermore, we address the utilization of KG embeddings in the Horizon2020 project Teaming.AI (www.​teamingai-project.​eu.) focusing on their respective benefits.

The fourth industrial revolution has driven companies of all sizes to embrace digitalization, recognizing the potential of AI technologies for data analysis and real-time decision-making. However, the adoption of AI by manufacturing SMEs faces challenges related to cost, accessibility, and the need for expertise. To address these challenges, this chapter introduces a groundbreaking platform developed as part of the EU H2020 KITT4SME project. The platform aims to democratize the adoption of AI tools by leveraging the “as-a-service” model, making them affordable and readily available for SMEs. It follows a five-step workflow (diagnose–compose–sense–intervene–evolve) to provide tailor-made AI solutions to SMEs. The distinctive functionality of the platform allows for the composition of AI components from a marketplace into a customized service offering for companies, filling a gap in existing AI platforms. The KITT4SME platform has been successfully applied in four use cases within the project and to 18 external demonstrators via Open Calls. This chapter presents one of the internal use cases to showcase the capabilities and benefits of the KITT4SME platform.

The development of B2B platforms has led to the diffusion of business models (BMs) based on the concept of sharing economy. In recent years, multisided platform BMs have become an important way of creating and capturing value even though the phenomenon remains undertheorized (Zhao et al., Long Range Planning 53(4):101892, 2020). Multisided platforms (MSPs) are present in an increasing number of sectors due to the development of the Internet, digital technologies, and artificial intelligence (AI). The manufacturing sector has not been untouched by this trend; however, it still struggles to establish value drivers to support small- and medium-sized enterprises (SMEs) in the change. The objective of the proposed study is to present an ecosystem for the SME manufacturing sector, which will be based on the selected MSP offering AI services. An initial BM for the AI platform as a service will be design, and a revenue model will be proposed within it. The selected case allowed the use of a methodological approach (PDT – Platform Design Toolkit) to the design of an MSP business model based on a qualitative analysis of the dynamics governing the platform ecosystem. The originality of the research stems from the reliance on data obtained from the implementation of the KITT4SME project (H2020, GA 952119). The study results indicate that it is crucial to properly identify the needs of the platform’s stakeholders, and then precisely define the values and the mechanisms for exchanging them through MSP.

Self-reconfiguration in manufacturing systems refers to the ability to autonomously execute changes in the production process to deal with variations in demand and production requirements while ensuring a high responsiveness level. Some advantages of these systems are their improved efficiency, flexibility, adaptability, and cost-effectiveness. Different approaches can be used for designing self-reconfigurable manufacturing systems, including computer simulation, data-driven methods, and artificial intelligence-based methods. To assess an artificial intelligence-based solution focused on self-reconfiguration of manufacturing enterprises, a pilot line was selected for implementing an automated machine learning method for finding and setting optimal parametrizations and a fuzzy system-inspired reconfigurator for improving the performance of the pilot line. Additionally, a deep learning segmentation model was integrated into the pilot line as part of a visual inspection module, enabling a more efficient management of the production line workflow. The results obtained demonstrate the potential of self-reconfigurable manufacturing systems to improve the efficiency and effectiveness of production processes.

Digital manufacturing tools aim to provide intelligent solutions that will help manufacturing industry adapt to the volatile work environment. Modern technologies such as artificial intelligence (AI) and digital twins (DT) are primarily exploited in a way to simulate and select efficient solutions from a broad range of alternative decisions. This work aims to couple DT and AI technologies in a framework where training, testing, and deployment of AI agents is made more efficient in production scheduling applications. A set of different AI agents were developed, utilizing key optimization technologies such as mathematical programming, deep learning, heuristic algorithms, and deep reinforcement learning are developed to address hard production schedule optimization problems. DT is the pilar technology, which is used to simulate accurately the production environment and allow the agents to reach higher efficiency. On top of that, Asset Administration Shell (AAS) technology, being the pilar components of Industry 4.0 (I4.0), was used for transferring data in a standardized format in order to provide interoperability within the multi-agent system (MAS) and compatibility with the rest of I4.0 ecosystem. The system validation was provided in the manufacturing system of the bicycle industry by improving the business performance.

Digital twin technology has become a driving force in the transformation of the manufacturing industry, playing a crucial role in optimizing processes, increasing productivity, and enhancing product quality. A digital twin (DT) is a digital representation of a physical entity or process, modeled to improve decision-making in a safe and cost-efficient environment. Digital twins (DTs) cover a range of problems in different domains at different phases in the lifecycle of a product or process. DTs have gained momentum due to their seamless integration with technologies such as IoT, machine learning algorithms, and analytics solutions. DTs can have different scopes in the manufacturing domain, including process level, system level, asset level, and component level. This work presents the knowlEdge Digital Twin Framework (DTF), a toolkit that comprises a set of tools to create specific instances of DTs in the manufacturing process. This chapter explains how the DTF relates to other standards, such as ISO 23247. This chapter also presents the implementation done for a dairy company.

The application of reinforcement learning often faces limitations due to the exploration phase, which can be costly and risky in various contexts. This is particularly evident in manufacturing industries, where the training phase of a reinforcement learning agent is constrained, resulting in suboptimal performance of developed strategies. To address this challenge, digital environments are typically created, allowing agents to freely explore the consequences of their actions in a controlled setting. Strategies developed in these digital environments can then be tested in real scenarios, and secondary training can be conducted using hybrid data that combines digital and real-world experiences.

In this chapter, we provide an introduction to reinforcement learning and showcase its application in two different manufacturing scenarios. Specifically, we focus on the woodworking and textile sectors, which are part of ongoing research activities within two distinct European Research Projects. We demonstrate how reinforcement learning is implemented in a digital context, with the ultimate goal of deploying these strategies in real systems.

To develop interoperable and flexible systems, Industry 4.0 solutions need available models and standardization. There is not just a need for physical but also digital asset descriptions, which can be reused in different cases. Particularly, with newer complex system integration, there is an increase in implementations of agent-based solutions. Although the meta-language Asset Administration Shell aims to help with integration, it is still not mature enough and lacks sufficient methods and tooling. To support the aim for interoperability, we present three tools: an updated generic agent model, a methodology for AAS model creation, and a platform for model visualization and distribution.

This chapter aims at presenting the system architecture of a distributed production testbed embedded in an interoperable Shared Production network. The goal of the modular architecture is to enable flexible, resilient and distributed production. The presented approach illustrates how Multi-Agent Systems (MAS) can be incorporated in the manufacturing domain for distributed components on different hierarchy layers based on a holonic approach. The concept is validated on the real-world demonstrator testbed of the SmartFactory^KL. Furthermore, the MAS is combined with Industry 4.0 technologies such as the Asset Administration Shell and OPC UA.

The manufacturing industry is an ecosystem full of changes and variations where production conditions are never the same. As an example, the raw materials received from suppliers differ from one another, though within tolerances. And similar variations appear in all areas of manufacturing, such as tool wearing, the statuses of production machines, and even operator decisions.

Improvements to production must factor in technical considerations and economic ones. Several points of view must be included to determine the best solution. Even when applying artificial intelligence (AI) in the manufacturing process, the situation should be similar: Several agents with different goals should interact to determine the most holistic solution.

This paper presents the ontology, semantics, and architecture that facilitates multiagent interaction. The Reference Architecture Model Industry 4.0, or RAMI 4.0 (RAMI4.​0 – 2018 – DE (plattform-i40.​de)), has been selected as the basis for this approach.

Given that most of the time, the operator’s decision is based on intuition and experience, not based on data analysis, this paper also analyses which data architecture will permit the data analysis of raw materials, finished products, tooling characteristics and statuses, machine parameters, and external conditions, to minimize the influence of intuition and personal bias on decision-making in manufacturing.

The labor market is a key part of an economy. Several existing online platforms allow the upload of resumes and the search for a job. One of their limitations, however, is that obtaining the best opportunity can be hard because certain jobs need some experiences, abilities, and features that an applicant might not know. The recent diffusion and employment of conversational agents definitely have proven to benefit this kind of issue. For example, ChatGPT has shown impressive outcomes in different domains and for a variety of tasks. It has weaknesses, although, related to the veracity of the responses it generates, which might deceive the user interacting with it. The usage of external domain knowledge is the direction we suggest in this chapter. Several lexical databases and taxonomies have already been collected and designed by different organizations. We illustrate a list of such resources and provide a solution that integrates conversational agents with relevant information extracted from one of such resources showing the benefits and the impact that our proposal can generate.

In today’s competitive landscape, networked production plays a crucial role in enabling companies to create value and remain competitive. By integrating advanced logistics and supply chain processes, companies optimise resources through cooperation and dynamic arrangements. However, managing the emerging complexity requires a new and intelligent approach. Decentralised Technical Intelligence (DTI) is a response to this challenge. It refers to the distributed and autonomous intelligence embedded in interconnected systems, devices, and agents—involving both humans and machines. By combining the strengths of humans and artificial intelligence (AI), DTI creates a coordinated environment that enhances the overall system intelligence. This collaboration leads to greater autonomy and enables multiple DTI agents to operate independently within a decentralised network. To achieve advanced networked production with DTI, a roadmap will be established, encompassing building blocks that focus on transparency, cooperation, sustainability, seamless integration and intelligent network control. All building blocks are linked to a vision, value promise and development pathway. As networked production evolves, it gives rise to new business models and demands new skills and expertise. By following this roadmap, DTI unlocks its potential for advancement, creating value and fostering competitiveness.

Improving worker safety and productivity is of paramount importance in the manufacturing industry, driving the adoption of advanced sensing and control systems. This concern is particularly relevant within the framework of Industry 5.0. In this context, wearable sensors offer a promising solution by enabling continuous and unobtrusive monitoring of workers’ activities in the manufacturing line. This book chapter focuses on wearable sensor-based human activity recognition and its role in promoting worker safety in manufacturing environments. Specifically, we present a case study on wearable sensor-based worker activity recognition in a manufacturing line with a mobile robot. As wearable sensors comprise various sensor types, we investigate and compare sensor data fusion approaches using neural network models to effectively handle the multimodal sensor data. In addition, we introduce several deep learning-based techniques to improve the performance of human activity recognition. By harnessing wearable sensors for human activity recognition, this book chapter provides valuable insights into improving worker safety on the manufacturing line, aligning with the principles of the Industry 5.0 paradigm. The chapter sheds light on the potential of wearable sensor technologies and offers avenues for future research in this field.

The primary goal of this research is to describe the scenarios, challenges, and complexities associated with object detection in industrial environments and to provide clues on how to tackle them. While object detection in production lines offers significant advantages, it also poses notable difficulties. This chapter delves into the common scenarios and specific challenges encountered in industrial object detection and proposes targeted solutions for various use cases. For example, synthetic data play a pivotal role in overcoming labeling challenges, particularly when it comes to small objects. By harnessing synthetic data, we can efficiently track and debug object detection results, ensuring faster identification and resolution of many data labeling issues. Synthetic data facilitate effective tracking and debugging of object detection results, streamlining the overall workflow. Furthermore, we explore the application of object detection in head-worn devices, utilizing the human point of view (POV) as a valuable perspective. This approach not only enhances human assistance systems but also enhances safety in specific use cases. Through this research endeavor, our aim is to contribute to the advancement of the whole process of object detection methods in complex industrial environments.

The adoption of AI in manufacturing enables numerous benefits that can significantly impact productivity, efficiency, and decision-making processes. AI algorithms can optimize production schedules, inventory management, and supply chain operations by analyzing historical data and producing demand forecasts. In spite of these benefits, some challenges such as integration, lack of data infrastructure and expertise, and resistance to change need to be addressed for the industry to successfully adopt AI. To overcome these issues, we introduce the AI Model Generation framework (AMG), able to automatically generate AI models that adjust to the user’s needs. More precisely, the model development process involves the execution of a whole chain of sub-processes, including data loading, automated data pre-processing, cost computation, automatic model hyperparameter tuning, training, inference, explainability generation, standardization, and containerization. We expect our approach to aid non-expert users into more effectively producing machine and deep learning algorithms and hyperparameter settings that are appropriate to solve their problems without sacrificing privacy and relying on third-party services and infrastructure as few as possible.

This chapter provides an introduction to common methods of anomaly detection, which is an important aspect of quality control in manufacturing. We give an overview of widely used statistical methods for detecting anomalies based on k-means, decision trees, and Support Vector Machines. In addition, we examine the more recent deep learning technique of autoencoders. We conclude our chapter with a case study from the EU project knowlEdge, where an autoencoder was utilized in order to detect anomalies in a manufacturing process of fuel tanks. Throughout the chapter, we emphasize the importance of humans-in-the-loop and provide an example of how AI can be used to improve manufacturing processes.

The industrial sector has been a major adopter of new technologies for decades, driving economic and societal progress. The path by which industry embraces new techniques has a significant impact on the environment and society and thus must be guided by principles of sustainability and trustworthiness. In this chapter, we explore the current paradigm in which Industry 4.0 is evolving towards Industry 5.0, where artificial intelligence (AI) and other advance technologies are being used to build services from a sustainable, human-centric, and resilient perspective. We examine how AI can be applied in industry while respecting trustworthy principles and collect information to define how well these principles are adopted. Furthermore, it is presented a perspective on the industry’s approach towards adopting trustworthy AI (TAI), and we propose steps to foster its adoption in an appropriate manner. We also examine the challenges and risks associated with the adoption of AI in industry and propose strategies to mitigate them. This chapter intends to serve researchers, practitioners, and policymakers interested in the intersection of AI, industry, and sustainability. It provides an overview of the latest developments in this field and offers practical guidance for those seeking to promote the adoption of TAI.

Industrial revolutions have historically disrupted manufacturing by introducing automation into production. Increasing automation reshapes the role of the human worker. Advances in robotics and artificial intelligence open new frontiers of human-machine collaboration. Such collaboration can be realized considering two sub-fields of artificial intelligence: active learning and explainable artificial intelligence. Active learning aims to devise strategies that help obtain data that allows machine learning algorithms to learn better. On the other hand, explainable artificial intelligence aims to make the machine learning models intelligible to the human person. The present work first describes Industry 5.0, human-machine collaboration, and state-of-the-art regarding quality inspection, emphasizing visual inspection. Then it outlines how human-machine collaboration could be realized and enhanced in visual inspection. Finally, some of the results obtained in the EU H2020 STAR project regarding visual inspection are shared, considering artificial intelligence, human-digital twins, and cybersecurity.

AI has gained significant traction in manufacturing, offering tremendous potential for enhancing production efficiency, cost reduction, and safety improvements. Consequently, developing AI-based software platforms that facilitate collaboration between human operators and AI services is crucial. However, integrating the different stakeholder perspectives into a common framework is a complex process that requires careful consideration. Our research has focused on identifying the individual relevance of varying quality characteristics per stakeholder toward such a software platform. Therefore, this work proposes an overview on the vital success factors related to human-AI teaming that can be used to measure fulfillment.

This chapter introduces the work performed in XMANAI to address the need of explainability in manufacturing AI systems applied to optimize production lines. The XMANAI platform is designed to meet the needs of manufacturing factories, offering them a unified framework to leverage their data and extract valuable insights. Within the project, the Ford use case is focused on forecasting production in a dynamically changing manufacturing line, serving as a practical illustration of the platform capabilities. This chapter focuses on the application of explainability using Hybrid Models and Heterogeneous Graph Machine Learning (ML) techniques. Hybrid Models combine traditional AI models with eXplainable AI (XAI) tools and Heterogeneous Graph ML techniques using Graph Attention (GAT) layers to extract explainability in complex manufacturing scenarios where data that can be represented as a graph. To understand explainability applied to the Ford use case, this chapter describes the initial needs of the scenario, the infrastructure behind the use case and the results obtained, showcasing the effectiveness of this approach, where models are trained in the XMANAI platform. Specifically, a description is given on the results of production forecasting in an engine assembly plant while providing interpretable explanations when deviations from expected are predicted.

Today, Explainable AI is gaining more and more traction due to its inherent added value to allow all involved stakeholders to understand why/how a decision has been made by an AI system. In this context, the problem of Product Demand Forecasting as faced by Whirlpool has been elaborated and tackled through an Explainable AI approach. The Explainable AI solution has been designed and delivered in the H2020 XMANAI project and is presented in detail in this chapter. The core XMANAI Platform has been used by data scientists to experiment with the data and configure Explainable AI pipelines, while a dedicated manufacturing application is addressed to business users that need to view and gain insights into product demand forecasts. The overall Explainable AI approach has been evaluated by the end users in Whirlpool. This chapter presents experiences and lessons learnt from this evaluation.

In today’s rapidly evolving technological landscape, businesses across various industries face a critical challenge: maintaining and enhancing the quality of both their processes and the products they deliver. Traditionally, this task has been tackled through manual analysis, statistical methods, and domain expertise. However, with the advent of artificial intelligence (AI) and machine learning, new opportunities have emerged to revolutionize quality optimization. This chapter explores the process and product quality optimization in a real industrial use case with the help of explainable artificial intelligence (XAI) techniques. While AI algorithms have proven their effectiveness in improving quality, one of the longstanding barriers to their widespread adoption has been the lack of interpretability and transparency in their decision-making processes. XAI addresses this concern by enabling human stakeholders to understand and trust the outcomes of AI models, thereby empowering them to make informed decisions and take effective actions.

The field of metrology, which focuses on the scientific study of measurement, is grappling with a significant challenge: predicting the measurement accuracy of sophisticated 3D scanning devices. These devices, though transformative for industries like manufacturing, construction, and archeology, often generate complex point cloud data that traditional machine learning models struggle to manage effectively. To address this problem, we proposed a PointNet-based model, designed inherently to navigate point cloud data complexities, thereby improving the accuracy of prediction for scanning devices’ measurement accuracy. Our model not only achieved superior performance in terms of mean absolute error (MAE) across all three axes (X, Y, Z) but also provided a visually intuitive means to understand errors through 3D deviation maps. These maps quantify and visualize the predicted and actual deviations, which enhance the model’s explainability as well. This level of explainability offers a transparent tool to stakeholders, assisting them in understanding the model’s decision-making process and ensuring its trustworthy deployment. Therefore, our proposed model offers significant value by elevating the level of precision, reliability, and explainability in any field that utilizes 3D scanning technology. It promises to mitigate costly measurement errors, enhance manufacturing precision, improve architectural designs, and preserve archeological artifacts with greater accuracy.