Fault diagnosis of electromechanical equipment faces three major challenges. At the signal level, the multidirectional and multimodal characteristics of equipment signals prevent a single sensor from fully characterizing equipment health states. At the engineering level, sharp variations in rotational speed induce domain shift phenomena, which further degrade diagnostic accuracy. At the application level, fused multi-source features provide strong representations but often exhibit limited generalizability and entail high deployment costs. In contrast, current signals are readily available and inexpensive to acquire, but their capability to represent mechanical faults is relatively weak. To address these challenges, this paper proposes the Synchronization Induction Network Based on Progressive Microelement Strategy and Hierarchical Dimensions (SI-CDN). SI-CDN incorporates a mutual-attention-driven information coupler that dynamically models correlations among signals collected at different measurement locations, thereby enabling collaborative representation learning from multiple signal sources. To mitigate the effects of rotational-speed variations, a progressive infinitesimal embedding and fractal reconstruction strategy is developed to iteratively refine feature representations through convolution and reconstruction operations. This strategy captures temporal patterns associated with rotational-speed variations while suppressing their adverse effects on feature extraction. Furthermore, an advanced similarity loss function is integrated into a cross-modal synchronization-induced distillation framework to dynamically transfer knowledge from multimodal fused features to current-signal features, thereby enhancing the representation capability of the current model. Experiments on two representative electromechanical systems demonstrate that SI-CDN achieves state-of-the-art fault diagnosis performance.
The growing volume of electronic waste poses a critical environmental burden and a significant challenge for resource circularity. It makes the efficient recovery of materials both a necessity and an opportunity, as the disassembly of these devices facilitates the recovery of hazardous and valuable materials, such as batteries and motors. The recovery process is usually constrained by the high variability in design, component layout, and internal structure across products in the same family. This is compounded by the actual state of the device, as parts can be missing or damaged. Such variability renders full automation intractable, compelling a reliance on manual or semi-automated operations. Technologies such as Mixed Reality can support operators by providing contextual instructions, such as the identification and localization of objects of interest. In complex products, the precision of the localization is important to correctly identify the target object for a given step and avoid identification errors. In many cases, product components are very close to each other and a localization error can lead to an identification mismatch. Although Artificial Intelligence-based 6D pose estimation models are key enablers for this functionality, they require large, accurately annotated training datasets, a labor-intensive bottleneck in industrial contexts. This paper proposes automating dataset annotation using foundation models. This automated approach, however, introduces inherent uncertainties regarding label precision that must be characterized. The primary contribution of this work is a rigorous, in-depth analysis of the sources of error propagating from automated dataset annotation. We evaluate the impact of these inaccuracies on model performance within the specific context of a real-world industrial disassembly application.
The transition toward Industry 5.0 promotes human-centred production systems in which digital technologies augment human capabilities rather than replace them. Within this paradigm, Digital Intelligent Assistants have emerged as promising tools to support operators during complex industrial tasks. Moreover, Generative Artificial Intelligence (GenAI), including Large Language Models and Vision Language Models, enables the development of such systems by combining natural language interaction with visual perception. This exploratory mixed-methods pilot study investigates the impact of a multimodal Digital Intelligent Assistant (M-DIA) designed to support operators during a manual assembly process. A controlled laboratory experiment involving 60 participants compared four conditions: multimodal interaction (voice and camera), voice-only interaction, camera-only interaction, and a manual baseline. The analysis considered process performance, perceived cognitive workload, usability and user experience, and operator behaviour. The results show that M-DIA improves assembly quality by reducing product errors compared with the manual condition, although no reduction in total process time is observed due to interaction-related delays. From a human perspective, the assistant reduces perceived cognitive workload, while higher frustration levels emerge in configurations involving voice interaction. Usability and user experience evaluations are overall positive across all experimental conditions. Behavioural observations further indicate that operators maintain an active supervisory role, critically assessing system outputs and avoiding systematic overreliance on the assistant. These findings provide preliminary empirical evidence on the role of GenAI-based multimodal assistants in human-centred manufacturing and highlight key design implications for their industrial deployment.
This paper proposes an autonomous material transportation system tailored for construction sites. The system comprises an unmanned material transport robot and a scheduling platform for coordinated control with other systems. A modular navigation planner is developed for point-to-point material transport tasks, enabling near-road navigation with limited deviation through integrated perception, global planning, local planning, and adaptive control. In addition, a rule-based hierarchical finite state machine is developed to execute the complete material transportation workflow. Compared with manual delivery, the proposed system achieves a 176.31% improvement in transportation efficiency. In terms of navigation performance, the planner demonstrates superior compliance with construction site specifications, along with enhanced planning stability and efficiency compared to conventional algorithms.
Goods-to-person order picking, such as robotic mobile fulfillment systems, is prevalent in e-commerce order fulfillment. However, the efficiency of robot transportation is hindered by the need to transport entire pods containing numerous stock-keeping units (SKUs), most of which are not required for the orders being fulfilled. This paper considers a new robotic cellular warehousing system that achieves efficient robot transportation by having robots transport only the SKUs requested in an order and by enabling them to perform both transport and picking operations. This is accomplished through a multi-robot parallel picking scheme, which involves three key stages of planning: (i) construction of order routes, (ii) multi-robot order allocation, and (iii) multi-robot conflict-free path planning. To facilitate the planning problems, we propose a dedicated time-space-state network model that considers three-dimensional information to identify feasible and efficient paths for robotic order picking. Our proposed approach is validated through computational experiments on realistic instances, and we also identify the optimal system configuration, including length-to-width ratio, robot fleet size, and their impacts on picking efficiency. The computational results and analyses offer practical managerial implications for warehouse operators, providing insights for decision-making.
