In the last few decades, several initiatives and approaches are set up to support maintenance procedures for the railway industry in adopting the principles of Industry 4.0. Contextualized maintenance technologies such as Augmented Reality (AR) overlay can integrate virtual information on physical objects to improve decision-making and action-taking processes. Operators work in a dynamic working environment requiring both high adaptive capabilities and expert knowledge. There is a need to support the operators with tailor-based information that is customized and contextualized to their expertise and experience. It calls for AR tools and approaches that combine complex methodologies with high usability requirements. The development of these AR tools could benefit from a structured approach. Therefore, the objective of this paper is to propose an adaptive architectural framework aimed at shaping and structuring the process that provides operators with tailored support when using an AR tool. Case study research is applied within a revelatory railway industry setting. It was found that the framework ensures that self-explanatory AR systems can capture the knowledge of the operator, support the operator during maintenance activities, conduct failure analysis, provide problem-solving strategies, and improve learning capabilities. This study contributes to the necessity of having a human-centered approach for the successful adaption of AR technology tools for the railway industry.

Maintenance and troubleshooting contribute to the overall economic efficiency of rolling stock throughout its lifecycle, yet these activities are resource-intensive and incur significant costs. In the era of Industry 4.0 implementation, the escalating complexity of rolling stock poses challenges in troubleshooting and maintenance decision-making. Recognizing this, there is a pressing need for a structured solution that empowers maintenance operators with tools for effective organization, analysis, and real-time presentation of rolling stock system failures, facilitating streamlined troubleshooting and maintenance guidance. Functional blocks, providing systematic structuring of processes, and Augmented Reality (AR), offering enhanced visualization and decision support, emerge as solutions to navigate and streamline these challenges in the evolving industrial landscape. The intricacies of existing fault-diagnosing systems, particularly interconnected transmission of functional blocks for troubleshooting and AR integration, lack detailed explanation, hindering effective solutions. To address these challenges, this work leverages AR for troubleshooting rolling stock failures, introducing functional blocks to (1) collect maintenance and fault data, (2) integrate maintenance information based on failure diagnostics and maintenance patterns, and (3) deliver this information to maintenance operators through AR. The developed Sequential Augmented Reality Assist (SARA) method, demonstrates the sequential implementation of AR troubleshooting functional blocks in an industrial context, providing a systematic solution to the complexities involved.

Industrial knowledge management (KM) remains highly fragmented, with knowledge capture, structuring, and application often treated as isolated activities. Conventional approaches depend on static documentation and experience-based practices, which limit their responsiveness in dynamic, operational environments. In this paper, we design a comprehensive framework for the integration of artificial intelligence (AI) and augmented reality (AR) in industrial maintenance and diagnostics. The framework leverages AI techniques, including natural language processing (NLP) for extracting and structuring domain knowledge and machine learning (ML) models for predictive fault classification. These AI capabilities are seamlessly combined with AR technologies to deliver immersive, context-aware, in-situ task guidance, thereby enhancing decision-making, reducing downtime, and supporting efficient, knowledge-based maintenance processes. A case study is employed to demonstrate the framework’s feasibility by developing a prototype system deployed in a maintenance setting. The AI module clusters maintenance actions and predicts task categories, while the AR component renders step-by-step maintenance instructions anchored in the physical workspace. This integration enables the transition from unstructured, digital maintenance records to step-by-step in-situ repair guidance, reducing reliance on expert memory or static manuals. The results show that combining AI-driven text analytics with AR-based visualisation creates a cohesive knowledge workflow that improves operational efficiency. The proposed framework offers a scalable approach to embedding KM into frontline industrial routines, laying the foundation for more adaptive, technician-centred knowledge systems.

The rapid evolution of Industry 4.0 and the convergence of Information Technology (IT) and Operational Technology (OT) have increased the complexity of industrial maintenance and troubleshooting. Cyber-Physical Production Systems (CPPS) generate large volumes of heterogeneous data, creating opportunities for predictive maintenance but also increasing the need for transparent and context-aware decision support. Existing maintenance solutions remain limited by fragmented workflows, black-box AI models, and Augmented Reality (AR) applications that primarily provide static procedural guidance without explainable reasoning. This paper presents NEXAR-Maint, a proof-of-concept framework that integrates neurosymbolic Artificial Intelligence (AI), Artificial Visual Intelligence (AVI), digital twins, Fault Tree Analysis (FTA), Root Cause Analysis (RCA), and AR to support explainable predictive maintenance and troubleshooting. The framework combines data-driven prediction with symbolic reasoning to provide context-aware diagnostics, while AR serves as the human–machine interface for visualising maintenance guidance and diagnostic pathways within the operator's workspace. To demonstrate the feasibility of the proposed architecture, an exploratory proof-of-concept user study involving three participants was conducted in a controlled maintenance troubleshooting scenario. The evaluation focused on the usability and feasibility of the explainable interaction concepts rather than on establishing statistically generalisable performance improvements. The results demonstrate the technical feasibility of integrating neurosymbolic AI, AVI, and AR into a unified maintenance framework and provide preliminary evidence that explainable AR-based decision support is perceived as useful by participants. These findings serve as a foundation for future large-scale industrial validation involving professional maintenance operators.

Tacit knowledge is crucial in industrial maintenance, yet it remains challenging to document and transfer through conventional training methods. This study addresses this by introducing the AR-COMPASS framework, which structures tacit expertise into six knowledge types using augmented reality (AR): collaborative routines, operational troubleshooting, motor skills and tool handling, practical shortcuts, adaptation to context, and safety awareness and risk assessment. Together with an overarching operator skill dimension that integrates the various knowledge types, this framework captures the embodied expertise that guides effective maintenance performance. To evaluate this framework, a case study was conducted in which an AR prototype was developed. The system integrates motion tracking, video documentation, and decision prompts to capture individual-level tacit knowledge. An expert mechanic's maintenance procedure was recorded, and the captured knowledge was embedded into the AR workflow. Subsequently, an A/B test was conducted with 10 participants to compare AR-guided instructions with a traditional paper-based manual. In the paper-based manual group, completion time improved by 9.4% and errors decreased by 60% between runs. The AR-guided group achieved substantially larger gains, with an 18.1% reduction in completion time and a complete elimination of critical errors in the second run. Participants also reported higher usability and confidence when using AR. These findings highlight the potential of the AR COMPASS framework to transform tacit knowledge into explicit, step-anchored guidance delivered at the point of need, providing a foundation for more reliable and digitally supported industrial maintenance.