Abstract Ergonomics evaluation methods are crucial for assessing risks for work-related musculoskeletal disorders and ensuring operator well-being, productivity, and safety. Despite the increased use of digital twins and AI-supported tools in production system design and operation, ergonomics evaluations still primarily rely on observational techniques such as expert assessments or checklist-based tools like the Rapid Entire Body Assessment and the Rapid Upper Limb Assessment. These methods are time-consuming, imprecise, and prone to subjectivity arising from variability in the judgment of the ergonomists and ambiguity in scoring criteria. As an alternative, ergonomics evaluation methods based on using technologies for direct measurements can provide semi-automation of the assessments and offer greater objectivity and precision. This study investigates the capability of a computer vision-based motion capture approach to support direct measurement ergonomics evaluations and compares its results with those of an inertial measurement unit-based system in an industrial task. The comparison was conducted by studying output data of the two systems and by feeding the data into a direct measurement-based ergonomics evaluation method. A representative industrial assembly task involving upper-body movement and dynamic wrist activity was recorded simultaneously using a single monocular RGB camera and an IMU-based system. Both datasets were processed using two parallel workflows that followed the same structure to extract joint angles and segment positions over time. The comparison and evaluation of the results demonstrates that computer vision-based motion capture has the potential to provide human posture and motion data suitable for direct measurement ergonomics evaluations in industrial environments.

Improving energy efficiency in industrial manufacturing remains challenging despite substantial technical potential. This has resulted in a persistent energy efficiency gap, which is increasingly understood as a socio-technical issue driven by not only technology limitations but also organizational and informational barriers. This study investigates how energy waste is targeted in practice through an in-depth single case study of an automotive company. Fifteen energy efficiency measures (EEMs) were analyzed and classified by type of energy waste addressed, digital technologies applied, and organizational knowledge required. The results show that industrial efforts primarily focus on reducing idling energy losses, while fewer measures address more complex forms of energy waste, such as over-processing losses. Digital technologies are mainly applied and rolled out at lower maturity levels, emphasizing energy monitoring and visualization. Further, different types of organizational knowledge are associated with targeting energy waste: technical knowledge dominates isolated interventions, process knowledge supports standardized technology diffusion, and leadership knowledge is required for cross-functional coordination. The findings highlight that bridging the energy efficiency gap requires the alignment of technological solutions with organizational knowledge and routines. This study contributes empirical insights into how manufacturing companies can structure and prioritize energy efficiency efforts and provides a framework to support the implementation of energy efficiency measures in practice.

This paper introduces Synthetic Active Learning (SAL), a fully automatic model refinement strategy for manufacturing parts detection using only synthetic data actively generated with domain randomization for training. SAL iteratively updates the detection model by identifying its weaknesses, such as in specific categories, materials, or object sizes, using custom evaluators, and generating targeted synthetic data to address them; it selectively synthesizes new useful data with respect to active learning, where traditionally humans in the loop select data to label. During each iteration, model training and data generation occur simultaneously to improve efficiency. Evaluated on four use cases from two industrial datasets, SAL achieved mAP@50 improvements of 2 to 6% percentage points over static learning, which refers to training on a fixed, pre-generated dataset. It also showed notable gains in underperforming categories, leading to more balanced performance across classes. Another benefit is that it uses a consistent configuration across multiple use cases, avoiding the need for extensive hyperparameter tuning common in prior domain randomization studies. Given its encouraging performance across diverse scenarios, we believe that SAL can scale to broader industrial applications where training can be fully or mostly based on synthetic data.

Digital human modelling (DHM) tools enable engineers to assess ergonomic conditions and predict risk factors before workstation designs are finalized, supporting proactive approaches to workplace safety. Despite these advantages, DHM tools remain underutilized in industry compared with computer aided design (CAD) and other digital engineering tools. Key documented barriers of the adoption of DHM tools include steep learning curves, requirements for specialized ergonomics expertise, time-consuming manual setup of simulations, and limited documentation resources. Meanwhile, advances in artificial intelligence (AI), particularly large language models (LLMs), offer new possibilities for supporting engineering workflows through natural-language interaction and automated analysis. However, structured approaches for integrating AI into DHM tools are largely absent. This paper proposes a framework for integrating AI capabilities into DHM tools and presents a prototype implementation to assess feasibility. The framework defines four functional roles: Dialog (natural-language interaction), Expert (documentation-grounded guidance), Analyser (ergonomics data interpretation), and Executor (command translation to simulation operations). The prototype, connected to a commercial DHM tool, demonstrates how AI-based support can guide users through modelling steps, automate repetitive tasks, and identify critical ergonomic indicators, illustrating how such support could potentially lower the threshold for non-expert users, reduce manual effort, and contribute to more systematic design of workstations that support worker well-being. The current work focuses on the conceptual framework and technical feasibility; formal user validation remains as future work.

This paper addresses key aspects of domain randomization in generating synthetic data for manufacturing object detection applications. To this end, we present a comprehensive data generation pipeline that reflects different factors: object characteristics, background, illumination, camera settings, and post-processing. We also introduce the Synthetic Industrial Parts Object Detection dataset (SIP15-OD) consisting of 15 objects from three industrial use cases under varying environments as a test bed for the study, while also employing an industrial dataset publicly available for robotic applications. In our experiments, we present more abundant results and insights into the feasibility as well as challenges of sim-to-real object detection. In particular, we identified material properties, rendering methods, post-processing, and distractors as important factors. Our method, leveraging these, achieves top performance on the public dataset with Yolov8 models trained exclusively on synthetic data; mAP@50 scores of 96.4% for the robotics dataset, and 94.1%, 99.5%, and 95.3% across three of the SIP15-OD use cases, respectively. The results showcase the effectiveness of the proposed domain randomization, potentially covering the distribution close to real data for the applications.

Usability is a key factor for successfully integrating new technology to aid an operator in production. It is measured using three metrics: efficiency (productivity), effectiveness (quality), and user satisfaction. One prominent technology for operator support is augmented reality (AR), which is mostly handheld or head-mounted. A human-centered approach is required to align the AR integration with the operator’s capabilities. The underlying use case in this study is an energy dashboard visualized using AR and non-AR media, namely, a monitor, tablet, and HoloLens. The resulting media applications were evaluated for usability in terms of efficiency, effectiveness, and satisfaction in the within-study experiments by 16 participants. Overall, the results showed increased efficiency and satisfaction for traditional-monitor users and increased effectiveness for tablet users. Despite the participants’ lack of experience with AR, the AR applications performed comparably to the monitor and even slightly better in some aspects. With the ongoing development of AR software and hardware, AR can become increasingly useful for machine monitoring in production. However, to use AR for more comprehensive tasks, its strengths and weaknesses must be considered.

Factory layout planning involves allocating resources and arranging equipment in manufacturing facilities to enhance system performance and ensure a safe work environment. Integrating digital human modeling tools into factory layout planning facilitates early worker well-being analysis, mitigating musculoskeletal disorders. This paper presents methods for modeling factory layout planning as a multi-objective reinforcement learning problem, leveraging digital human modeling-based simulations.

This paper explores the use of digital technology stages and knowledge demand types for achieving energy efficiency. Digital technology stages are the steps toward developing an intelligent and networked factory: computerization, connectivity, visibility, transparency, predictive capacity, and adaptability. Knowledge demand types refer to the knowledge and skills needed to implement energy management through technical, process, and leadership knowledge. Empirical data were collected from a critical single case study at an industrial manufacturing company. The study made two significant contributions. Firstly, it identifies fourteen challenges and improvement potentials when working with energy monitoring, evaluation, and optimization, demonstrating the critical role of digital technology stages and knowledge demand types. Secondly, the study presents a conceptual framework indicating how companies could overcome pitfalls and enhance energy efficiency by combining digital technologies and knowledge demands. Future work will include technical implementations and its connection to knowledge management.

Rising energy costs, energy supply uncertainties, and the sustainability crisis have intensified the need for energy efficiency in industrial manufacturing. This adds complexity to balancing traditional production goals such as productivity, quality, and cost. While prior studies address energy-intensive processes or throughput bottlenecks, they often lack integrated decision-support for evaluating optimal trade-offs. To address this gap, this study proposes a novel simulation-based multi-objective optimization framework combined with a knowledge discovery module, demonstrated in an industrial case study. The framework systematically identifies energy and productivity losses, evaluates improvement strategies to determine optimal trade-off solutions, and extracts actionable rules to guide decision making. Case study results show a 23.9% reduction in specific energy consumption and a 27.9% increase in throughput, while emphasizing the need to balance inventory levels. The approach offers a robust, data-driven method for supporting energy-efficient manufacturing. Future research will explore integration with real-time monitoring and extension to additional objectives such as costs and emissions.

Industry 5.0 places worker's wellbeing at the center of the production process, prioritizing healthy and safety job conditions. Requirements to achieve occupational wellbeing are reducing risks for Work-related Musculoskeletal Disorders (WMSDs) and improving industry workstations. The traditional ergonomics risk assessments are based on human observational evaluation and the results are influenced by observers' competence. Nowadays, advanced technologies such as motion capture systems are implemented to objectively monitor an operator's movements over time. By providing real-time, data-driven insights into human movement and posture, systems offer the potential to reduce workplace injuries, enhance productivity, and promote long-term worker health. The purpose of the present study is to evaluate and compare three different approaches for assessing the quantitative biomechanical risk of an industrial task using the RULA method: the observational method, a wearable inertial measurement system, and a 4D stereophotogrammetry. The experiment involves one participant (female, 30 years old) performing a "pick-and-place" worker's task in a controlled laboratory environment. RULA scores vary across the three approaches, with discrepancies primarily due to differences in how each system captures and measures joint angles. While this preliminary study provides valuable initial insights, the limitation of involving a single participant must be critically acknowledged. Future research will aim to include a larger sample size and conduct statistical analyses. The identification of benefits and limitations of each approach enables researchers, ergonomists, and industry stakeholders to critically select and integrate technology to support the worker's safety, optimizing human wellbeing and overall system performance.

Assembly quality inspection plays a vital role in manufacturing, where correct part placement and alignment directly affect product reliability. While deep learning-based object detection offers a promising solution for automatic assembly quality inspection, it is hindered by data scarcity. Training on synthetic data with Domain Randomization (DR) helps address this challenge, yet existing DR methods focus on generating individual objects and do not capture the relational structure needed for assembly inspection. In this paper, we identify two key factors for effective synthetic data generation in assembly inspection: preserving spatial relationships between components and providing part-level textures and annotations. We propose an Assembly-Specific Generation Scheme that incorporates these factors into a state-of-the-art DR pipeline. To evaluate its impact, we introduce SIP2A-OD, a new object detection dataset comprising two real-world assembly use cases collected under varied manufacturing conditions. We train a YOLOv12 model on synthetic data generated by our pipeline and test it on real data from the SIP2A-OD dataset. Compared to the baseline pipeline designed for individual object detection, our method improves mAP@50 by more than 15% in both use cases. These results demonstrate the effectiveness of our scheme and its potential for broader applications in industrial assembly inspection without the need for manual data collection or annotation.

Planning and designing factory layouts are frequently performed within virtual environments, relying on inputs from various cross-disciplinary activities e.g., product development, manufacturing process planning, resource descriptions, ergonomics, and safety. The success of this process heavily relies on the expertise of the practitioners performing the task. Studies have shown that layout planning often hinges on the practitioners’ knowledge and interpretation of current rules and requirements. As there is significant variability in this knowledge and interpretation, there is a risk that decisions are made on incorrect grounds. Consequently, the layout planning process depends on individual proficiency. In alignment with Industry 4.0 and Industry 5.0 principles, there is a growing emphasis on providing practitioners involved in industrial development processes with efficient decision support tools. This paper presents a digital support function integrated into a virtual layout planning tool, developed to support practitioners in considering current rules and requirements for space claims in the layout planning process. This digital support function was evaluated by industry practitioners and stakeholders involved in the factory layout planning process. This initiative forms part of a broader strategy to provide advanced digital support to layout planners, enhancing objectivity and efficiency in the layout planning process while bridging cross-disciplinary gaps.

Recently the concept of Industry 5.0 has been introduced, reinforcing the human-centric perspective for future industry. The human-centric scientific discipline and profession ergonomics is applied in industry to find solutions that are optimized in regard to both human well-being and overall system performance. It is found, however, that most production development and preparation work carried out in industry tends to address one of these two domains at a time, in a sequential process, typically making optimization slow and complicated. The aim of this paper is to suggest, demonstrate, and evaluate a concept that makes it possible to optimize aspects of human well-being and overall system performance in an efficient and easy parallel process. The concept enables production planning and balancing of human work in terms of two parameters: assembly time as a parameter of productivity (system performance), and risk of musculoskeletal disorders as a parameter of human well-being. A software demonstrator was developed, and results from thirteen test subjects were compared with the traditional sequential way of working. The findings show that the suggested relatively unique parallel approach has a positive impact on the expected musculoskeletal risk and does not necessarily negatively affect productivity, in terms of cycle time and time balance between assembly stations. The time to perform the more complex two-parameter optimization in parallel was shorter than the time in the sequential process. The majority of the subjects stated that they preferred the parallel way of working compared to the traditional serial way of working.

The convergence of the focus of Industry 5.0 on human well-being and the prevalent problem of work-related musculoskeletal disorders necessitates advanced digital solutions due to limitations in manual risk assessment methods. This research aimed to compare usability of a newly developed video-based awkward posture identification software, the ergonomist assistant for evaluation (ERAIVA) with a conventional manual method. The risk assessment tool utilised in this study, integrated into the ERAIVA digital platform, is the risk management assessment tool for manual handling proactively (RAMP). Four assessors evaluated video-recorded tasks using both methods (manual and ERAIVA). The usability was assessed through the post-study system usability questionnaire, time consumption, number of video replays and video annotation deletions. The impact on identification of awkward posture durations was also studied. ERAIVA exhibited the highest usability score; it showed a higher number of video replays of specific sequences and annotations without significant differences in time consumption.

This paper presents a decision support approach to enable decision-makers to identify no-preference solutions in multi-objective optimization for factory layout planning. Using a set of trade-off solutions for a battery production assembly station, a decision support method is introduced to select three solutions that balance all conflicting objectives, namely, the solution closest to the ideal point, the solution furthest from the nadir point, and the one that is best performing along the ideal nadir vector. To further support decision-making, additional analyses of system performance and worker well-being metrics are integrated. This approach emphasizes balancing operational efficiency with human-centric design, aligning with human factors and ergonomics (HFE) principles and Industry 4.0–5.0. The findings demonstrate that objective decision support based on Pareto front analysis can effectively guide stakeholders in selecting optimal solutions that enhance both system performance and worker well-being. Future work could explore applying this framework with alternative multi-objective optimization algorithms.

In the manufacturing industry, automatic quality inspections can lead to improved product quality and productivity. Deep learning-based computer vision technologies, with their superior performance in many applications, can be a possible solution for automatic quality inspections. However, collecting a large amount of annotated training data for deep learning is expensive and time-consuming, especially for processes involving various products and human activities such as assembly. To address this challenge, we propose a method for automated assembly quality inspection using synthetic data generated from computer-aided design (CAD) models. The method involves two steps: automatic data generation and model implementation. In the first step, we generate synthetic data in two formats: two-dimensional (2D) images and three-dimensional (3D) point clouds. In the second step, we apply different state-of-the-art deep learning approaches to the data for quality inspection, including unsupervised domain adaptation, i.e., a method of adapting models across different data distributions, and transfer learning, which transfers knowledge between related tasks. We evaluate the methods in a case study of pedal car front-wheel assembly quality inspection to identify the possible optimal approach for assembly quality inspection. Our results show that the method using Transfer Learning on 2D synthetic images achieves superior performance compared with others. Specifically, it attained 95% accuracy through fine-tuning with only five annotated real images per class. With promising results, our method may be suggested for other similar quality inspection use cases. By utilizing synthetic CAD data, our method reduces the need for manual data collection and annotation. Furthermore, our method performs well on test data with different backgrounds, making it suitable for different manufacturing environments.

Enhancing data visualization on the shop floor provides support for dealing with the increasing complexity of production and the need for progressing towards emerging goals like energy efficiency. It enables personnel to make informed decisions based on real-time data displayed on user-friendly interfaces. Augmented reality (AR) technology provides a promising solution to this problem by allowing for the visualization of data in a more immersive and interactive way. The aim of this study is to present a framework to visualize live and historic data about energy consumption in AR, using Power BI and Unity, and discuss the applications’ capabilities. The study demonstrated that both Power BI and Unity can effectively visualize near-real-time machine data with the aid of appropriate data pipelines. While both applications have their respective strengths and limitations, they can support informed decision-making and proactive measures to improve energy utilization. Additional research is needed to examine the correlation between energy consumption and production dynamics, as well as to assess the user-friendliness of the data presentation for effective decision-making support.

In line with Industry 5.0, ergonomic factors have recently received more attention in balancing assembly lines to enhance the human-centric aspect. Meanwhile, today’s mass-customized trend yields manufacturers to offset the assembly lines for different product variants. Thus, this study addresses the mixed-model assembly line balancing problem (MMALBP) by considering worker posture. Digital human modeling and posture assessment technologies are utilized to assess the risks of work-related musculoskeletal disorders using a method known as rapid entire body analysis (REBA). The resulting MMALBP is formulated as a mixed-integer linear programming (MILP) model while considering three objectives: cycle time, maximum ergonomic risk of workstations, and total ergonomic risks. An enhanced non-dominated sorting genetic algorithm (E-NSGA-II) is developed by incorporating a local search procedure that generates neighborhood solutions and a multi-criteria decision-making mechanism that ensures the selection of promising solutions. The E-NSGA-II is benchmarked against Epsilon-constraint, MOGA, and NSGA-II while solving a case study and also test problems taken from the literature. The computational results show that E-NSGA-II can find promising Pareto front solutions while dominating the considered methods in terms of performance metrics. The robustness of E-NSGA-II results is evaluated through one-way ANOVA statistical tests. The analysis of results shows that a smooth distribution of time and ergonomic loads among the workstations can be achieved when all three objectives are simultaneously considered.

This paper is about effectively utilizing synthetic data for training deep neural networks for industrial parts classification, in particular, by taking into account the domain gap against real-world images. To this end, we introduce a synthetic dataset that may serve as a preliminary testbed for the Sim-to-Real challenge; it contains 17 objects of six industrial use cases, including isolated and assembled parts. A few subsets of objects exhibit large similarities in shape and albedo for reflecting challenging cases of industrial parts. All the sample images come with and without random backgrounds and post-processing for evaluating the importance of domain randomization. We call it Synthetic Industrial Parts dataset (SIP-17). We study the usefulness of SIP-17 through benchmarking the performance of five state-of-the-art deep network models, supervised and self-supervised, trained only on the synthetic data while testing them on real data. By analyzing the results, we deduce some insights on the feasibility and challenges of using synthetic data for industrial parts classification and for further developing larger-scale synthetic datasets. Our dataset † and code ‡ are publicly available.