Over the past three years, we collaborated with Swedish industries to explore how Augmented Reality (AR) can support production workflows in industrial manufacturing environments. This paper reflects on the design, development, and testing of AR prototypes for head-mounted optical see-through displays in real-world industrial settings. We investigate AR deployment across four manufacturing contexts: house production, inspection, tool change, and changeover operations. Through iterative prototyping and cross-case analysis, we identify recurring usability and integration challenges and extract corresponding user needs. These insights inform a set of design guidelines for AR in industrial production. 

The requirements for rapidly changing product designs and product customisation increase the need for manufacturing systems to be updated frequently. Existing approaches, such as Plug & Produce, use standardised resources that can be moved around quickly when needed to adapt to these requirements. However, robot controllers still require reprogramming or extensive reconfiguration to work with the new setups. The use of a Large Language Model (LLM) has potential in assisting in automatically adapting these systems to new product designs. When changes are frequent, the use of in-house knowledge should be the focus rather than the use of external expert knowledge. Using LLM, a manufacturing system can be instructed on what to do using natural language. This simplifies and speeds up the changes needed to adapt to new product requirements. This article presents an implemented and tested system for using LLM with a physical collaborative robot equipped with a mechanical gripper and a vision system. This is tested with a kitting application that includes a set of buffers for holding the objects to be kitted, a human giving instructions to the system, and a kitting tray to hold the kit.

The manufacturing industry faces a shortage of skilled workers and an increasing demand for customized products, making traditional automation unsuitable for the emerging high-mix, low-volume production. There is a growing need for intelligent automation that enables in-house knowledge to perform frequent reconfigurations. A promising solution involves distributed intelligence, where products autonomously manage their goals, and modular resources possess the necessary skills to achieve them. Process planners aim to digitally configure manufacturing processes and sequencing without relying on detailed, device-specific programming. This article validates a configurable multi-agent-based control system that employs Artificial Intelligence-driven transport planning using Satisfiability Modulo Theories (SMT), ensuring cost-effective adaptation to new parts, materials, and resources. A Plug & Produce robot station, equipped with tools, tool holders, and related components for a kitting application, is utilized for validation. This study compares automatically planned tool changes with the complexity of traditional Programmable Logic Controller (PLC) and robot programming in on-demand manufacturing.

This article presents a system for automatically updating a digital twin model, used for automated path planning of an industrial robot. The digital twin needs to be accurately calibrated in relation to the resource locations due to the physical limitations of placing resources out precisely. The process considered is a surface roughness measurement of aerospace metal parts that requires high positional accuracy.

The scenario takes place in a robot cell that is a Plug & Produce system, where resources can be added and removed in minutes, allowing fast reconfiguration of the production resources. This means that an automated path planner is required for the robot to adapt to new locations of these resources automatically. A digital twin is proposed, consisting of a robot path planner and a simulation model that is updated when resources are added to the system. The resources should automatically appear in the simulation and be placed at an accurate location.

The purpose of automating these steps is to make the update of the digital twin faster during production and remove the requirement for expert knowledge.

The study contributes to how transdisciplinary research collaboration can be designed to address the complexity of the human-technology nexus in the context of Industry 5.0 and to identify incentives from the manufacturing industry to engage in transdisciplinary research efforts. Engaged scholarship and work-integrated learning approaches are applied to integrate diverse research disciplines and active stakeholder engagement to address complex societal challenges. The methodology of this research is a qualitative case study, including workshops and focus groups with a project consortium of eight companies, with industry experts and university researchers. Findings contribute to transdisciplinary research collaboration viewed as an iterative continuous process including three phases for the process of reaching full potential of transcending disciplines and organizations. Contribution shows that industry highlights the need to address human challenges in smart technology adoption, motivating engagement in transdisciplinary research. Advancing smart manufacturing requires embracing creativity and innovation in the human-technology nexus. Further, transdisciplinary research collaboration needs to be based on trust, relationships, sharing, courage, mutual understanding and respect for each other's disciplines and expertise. The collaborative design accentuates the significance of transdisciplinary research in university-industry collaboration when moving forward with human-centric and smart manufacturing in line with the evolving Industry 5.0 paradigm

This article delves into the application of Virtual Commissioning (VC) using Digital Twin (DT) technology to optimize automated manufacturing systems. The research focuses on a vane scanning process within an industrial test environment. The main goal is to identify the challenges of integrating VC and DT into current manufacturing workflows, with a strong emphasis on early fault detection, optimizing systems, and efficient use of resources. This article provides a structured framework for VC, offering strategic information to make the VC implementation process easier and more efficient for industry practitioners. In addition to that, it significantly contributes to understanding VC’s potential to make manufacturing processes more efficient, lower operational costs, and improve overall system performance. A key outcome of this research was the successful validation of a virtual model, which played a crucial role in identifying essential factors for effective VC implementation. These factors include ensuring tool compatibility, maintaining interoperability, and integrating real-time data during the commissioning process. Furthermore, the findings are particularly valuable for advancing smart manufacturing practices and enhancing the deployment of automated systems as well as the importance of incorporating VC into the development and procurement of new automated manufacturing systems, providing a clear path to fully utilize the benefits of digital twin technology in industrial settings.  

Modern smart manufacturing requires flexible and safe coordination between human operators and autonomous agents. Current multi-agent system planning often relies on reactive safety mechanisms, leading to inefficiencies and workflow interruptions. Existing approaches typically overlook runtime safety policies during plan generation, producing functionally valid but operationally inefficient plans. As a result, avoidable slowdowns and emergency stops occur when humans enter shared workspaces, reducing both throughput and predictability.

This research introduces a safety-aware planning approach that builds on an enhanced agent ontology by integrating agent negotiation with automated planning. A dynamic map of enhanced edges is constructed, where each edge represents a sequence of skills and its aggregated runtime execution cost. During runtime, agents negotiate to exclude unsafe edges and update costs based on runtime conditions, enabling the solver to generate plans that are both safe and efficient. Evaluation in a Plug & Produce case study with two system configurations demonstrated consistent advantages over a baseline reactive safety policy. The proposed planner, increased throughput by 50-80%, reduced execution costs by 20-55%, and lowered variability by up to a factor of 17. By integrating safety awareness into the planning stage, the approach improves efficiency, robustness, and compliance with human safety standards while supporting adaptable reconfigurable manufacturing.

This work advances the vision of a human-centric future manufacturing by demonstrating that smart manufacturing systems can reason about operator presence to avoid hazards without compromising productivity.

Adapting automated manufacturing to accommodate new product designs is often challenging in traditional manufacturing systems. Plug & Produce is a concept that enables for faster adaptation to new product requirements. Research has shown that this can be done through standardized, modular resources that can be easily reconfigured and moved around. However, new flexible approaches are needed to make these modules work together with the automation control system. Typically, these changes to automation requires re-programming or advanced reconfigurations. This article presents an approach using Large Language Models to simplify the steps to instruct a Plug & Produce system on what to do.

The aim is to make use of inhouse knowledge, rather than external experts for adapting the manufacturing system to new product designs. It is identified that simulations are important tools for evaluating the generated instructions before deploying them to a physical system. The proposed system is implemented and the result in this article shows that automated manufacturing can be adapted by using a natural language.  

The demand for customized products in a saturated market of trendy customers forces the manufacturing industries to transform their manufacturing from a high volume of uniformed products to low volumes and a high mix of products. High mix and low volume manufacturing is most often manually performed since existing automation solutions are only profitable for mass manufacturing, due to explicitly designed control software where the product data is implemented as low-level control code. Highly flexible automated manufacturing systems such as Plug & Produce are requested, but challenges still exist before industrial implementation. This article proposes a digitally configurable system where data for new or modified products data is configured from the perspective of the product and its manufacturing processes instead of the manufacturing resources. In a Plug & Produce system, process modules with manufacturing resources are easy to replace for new or modified products and possibly to duplicate if higher capacity is needed. Configurable multi-agent systems are proposed by several researchers as a control system for Plug & Produce. An agent is a piece of autonomous computer code that negotiates with other agents and concurrently solves tasks, distributed on parts and resources. A part is a part of a product and part agents handle manufacturing goals for the parts. Resource agents know their capability and start operating as soon as they are plugged in. Resource agents follow pluggable process modules containing manufacturing resources and act as drivers for the modules. Gantry robots have by design a naturally orthogonal coordinate system and most often lack the functionality to handle work and tool coordinate objects as standard industrial robots do. Work objects refer to a base coordinate system and tool objects contain a reference to the tool center point. These references are in this article integrated into resource agents together. A place coordinate agent has the global perspective of the Plug & Produce cell and provides the process modules with reference coordinates of the place they are plugged into. Coordinates are recalculated from a product perspective into a resource perspective by coordinate transformations built into the skills of resource agents. This structure enables the possibility for process planners in the manufacturing company to make changes on a daily basis. A test with a gantry robot Plug & Produce demonstrator was performed and presented in this article to verify the generic structure of the gantry robot control system into agents. 

Industry 5.0, which focuses on human-centric automation and utilizes advanced production technologies such as Reconfigurable Manufacturing Systems (RMS), requires manufacturers to prioritize workers’ well-being alongside efficiency. Addressing safety management in this evolving manufacturing paradigm is essential. However, ensuring safety in reconfigurable manufacturing often requires external outsourcing and increased man-hours. This leads to increased production costs and reduced flexibility due to the additional time required for safety assurance. Ideally, manufacturers seek safety management methods that leverage in-house expertise, reducing both production costs and time without compromising safety. Thus, a novel approach to safety management is necessary. This paper introduces a method for software-assisted safety management in RMS that leverages in-house competencies and streamlines safety validation after reconfiguration which enhances Industry 5.0’s adaptability. To empirically assess the proposed method, a conceptual software tool was developed and deployed to a reconfigurable Plug & Produce system for house wall fabrication within a laboratory setting. A usability test was performed to collect the man-hour needed for safety validation after reconfiguration using in-house competency. Analysis of the results revealed potential savings of 40% for one-off production and 35% for batches up to 5. While based on lab findings, they suggest cost reduction in real manufacturing. This empirical evidence underscores significant cost reduction potential in reconfigurable manufacturing, highlighting its role in promoting flexibility, economical sustainability, and human-centricity within Industry 5.0 © 2024 IEEE.