The integration of Artificial Intelligence (AI) into manufacturing is transforming the industry by advancing predictive maintenance, quality control, and supply chain optimisation, while also driving the shift from Industry 4.0 towards a more human-centric and sustainable vision. This emerging paradigm, known as Industry 5.0, emphasises resilience, ethical innovation, and the symbiosis between humans and intelligent systems, with AI playing a central enabling role. However, challenges such as the “black box” nature of AI models, data biases, ethical concerns, and the lack of robust frameworks for trustworthiness hinder its widespread adoption. This paper provides a comprehensive survey of AI trustworthiness in the manufacturing industry, examining the evolution of industrial paradigms, identifying key barriers to AI adoption, and examining principles such as transparency, fairness, robustness, and accountability. It offers a detailed summary of existing toolkits and methodologies for explainability, bias mitigation, and robustness, which are essential for fostering trust in AI systems. Additionally, this paper examines challenges throughout the AI pipeline, from data collection to model deployment, and concludes with recommendations and research questions aimed at addressing these issues. By offering actionable insights, this study aims to guide researchers, practitioners, and policymakers in developing ethical and reliable AI systems that align with the principles of Industry 5.0, ensuring both technological advancement and societal value.

Artificial intelligence (AI) is increasingly adopted in manufacturing for tasks such as automated inspection, predictive maintenance, and condition monitoring. However, the opaque, black-box nature of many AI models remains a major barrier to industrial trust, acceptance, and regulatory compliance. This study investigates how explainable artificial intelligence (XAI) techniques can be used to systematically open and interpret the internal reasoning of AI systems commonly deployed in manufacturing, rather than to optimise or compare model performance. A unified explainability-centred framework is proposed and applied across three representative manufacturing use cases encompassing heterogeneous data modalities and learning paradigms: vision-based classification of casting defects, vision-based localisation of metal surface defects, and unsupervised acoustic anomaly detection for machine condition monitoring. Diverse models are intentionally employed as representative black-box decision-makers to evaluate whether XAI methods can provide consistent, physically meaningful explanations independent of model architecture, task formulation, or supervision strategy. A range of established XAI techniques, including Grad-CAM, Integrated Gradients, Saliency Maps, Occlusion Sensitivity, and SHAP, are applied to expose model attention, feature relevance, and decision drivers across visual and acoustic domains. The results demonstrate that XAI enables alignment between model behaviour and physically interpretable defect and fault mechanisms, supporting transparent, auditable, and human-interpretable decision-making. By positioning explainability as a core operational requirement rather than a post hoc visual aid, this work contributes a cross-modal framework for trustworthy AI in manufacturing, aligned with Industry 5.0 principles, human-in-the-loop oversight, and emerging expectations for transparent and accountable industrial AI systems.

Enterprise Architecture (EA) is crucial in any organisation as it defines the basic building blocks of a business. It is typically presented as a set of documents that help all departments understand the business model. In EA, safety documents are used to manage and understand safety risks. A novel similarity system for railway safety document processing is presented in this work. It measures the feasibility of automated updating of EA models with the Rule Book by verifying whether Rail Safety and Standards Board (RSSB’s) Rule Book clauses are present and complete in existing EA models. Additionally, a Natural Language Processing (NLP) based search feature was developed to drill through the database to find similar existing rules, principles, and clauses based on semantic similarity. The result will display the most similar clauses and rules with similarity scores and document names. In this study, different pre-trained Electra Small, DistilBERT (Distillation Bidirectional Encoder Representations from Transformers) Base and BERT (Bidirectional Encoder Representations from Transformers) Base were used to embed text. Additionally, the similarity between document rules was measured by cosine similarity metrics. With conclusive evidence, our findings show that BERT Base exceeds the other embedding methods in the semantic comparison of documents.

This work is concerned with a comparison between commercial available high-power LEDs, which have the same chip size but different power ratings (20 W, 30 W and 50 W), to find the most appropriate for Li-Fi applications, in terms of bandwidth before and after bandwidth extension. Two basic types of commercial LEDs were compared (`cool' and `warm'), where the bandwidth of both types was in the range of 3 to 4 MHz, for all power values. After bandwidth extension, there was a marked disparity in the bandwidth between the several types, where the bandwidth increased to approximately 17 MHz for the `warm', 33 MHz for the 50 W `cool' and 36 MHz for 20 W and 30 W `cool' LEDs. The bandwidth extended LEDs were implemented for a Li-Fi application based on offset PPM modulation transmission scheme at a rate of 13 Mbps. It was found that `cool' LEDs have a more favourable performance than `warm' LEDs in terms of data transmission and bit error rate (BER). Additionally, 20 W and 30 W LEDs have achieved a lower BER than the 50 W.

In this paper a dicode pulse position modulation (DiPPM) technique has been successfully implemented for an indoor visible light communication (VLC) based system using an FPGA and a commercial high power white LED (30 W). A data rate of 13 Mbps has been achieved with a bit error rate (BER) <;10 ^-11 at a free space distance of 1.2 m through a basic transceiver circuit, and 14.5 Mbps with 1.15×10 ^-6 of BER. Furthermore, a data rate of 13 Mbps has been successfully transmitted over a distance of 1.8 m with BER of 10 ^-5 . This proves that the application of DiPPM is a viable modulation system over free space communication links.

Low bandwidth expansion modulation schemes are preferred for free space and optical fibre data transmission, where limited bandwidth is available. One such scheme is duobinary pulse position modulation (DuoPPM), which is the subject of this paper. DuoPPM scheme is not as sensitive to bandwidth expansion issues as digital PPM, with a line rate of twice the data rate. This paper discusses first time practical implementation of DuoPPM coding scheme and its application in free space using visible light LED (30 W) for transmission purposes. Experimental results achieved at the data rate of 14 Mbit/s indicate an error rate that is better than 1 error in 10⁹. The main aim is to analyse the practicality, robustness and limitations of DuoPPM.