In order to obtain high-quality innovative concepts during the design process, designers often need to fusion multidisciplinary knowledge. However, cognitive fixation among designers and the technical distance between domains make it difficult to effectively apply cross-domain knowledge. To address these issues, this study proposes a C-K theory-driven systematic method for patent-based cross-domain knowledge fusion, aiming to enhance the feasibility of applying cross-domain knowledge in designer-led innovative design processes. First, a C&K space representation model is constructed to activate associations between concepts from different domains through seven types of conceptual relationships, enabling multi-level expression of cross-domain knowledge based on knowledge attributes. Second, knowledge intermediaries with different roles are selected based on the similarity of C&K space resources between the source and recommended domains, establishing corresponding retrieval paradigms to acquire intermediary domain knowledge. Then, a link prediction method is employed to construct an intermediary knowledge interaction network, recommending optimal cross-domain knowledge fusion paths from an attribute perspective to assist designers in overcoming decision paralysis in complex solution spaces. On this basis, a systematic cross-domain knowledge fusion innovation design process model is established, and the effectiveness of the method is verified using a garlic peeling machine as a case study.
Under the paradigm of mass personalisation, enterprises are compelled to accurately and efficiently extract functional requirements (FRs) to respond to ever-changing customer requirements (CRs) and evolving technology-driven requirements (TDRs). However, traditional FRs mining methods suffer from high labour intensity, fragmented design knowledge, and insufficient integration of heterogeneous requirement data. To address these issues, an FRs mining approach based on a heterogeneous data-driven knowledge graph (HDDKG) is proposed, in which the CRs domain, TDRs domain and FRs domain are integrated into a unified multi-domain representation. Firstly, a quadruple-task CRs extraction and classification model is developed to capture both explicit and implicit CRs from online reviews, while a tailored relation extraction model is constructed to identify TDRs and their functional semantics from patent documents. Secondly, FR entities are extracted from product instructions. Subsequently, a Cosine Sentence model is employed to establish semantic mappings from CRs and TDRs to FRs. Finally, a case study on FRs mining is conducted to verify the effectiveness of the proposed HDDKG-based approach, and an extended exploration of the LLM-based multi-agent system provides a promising direction for future intelligent FRs mining.
Resilient manufacturing plays a key role in Industry 5.0, focusing on personalised demands and emphasising the combination of production flexibility and smart technologies. It enables workshops to adjust production more flexible to meet dynamic production needs, while optimising manufacturing processes. Autonomous Mobile Robots (AMRs) have become crucial in enhancing resilient manufacturing by dynamically adapting to real-time material handling tasks. Based on multi-agent communication, this paper proposes a distributed collaborative scheduling scheme comprising the autonomous perception, collaborative interaction, and scheduling decision aimed at enhancing workshop material handling resilience. First, based on the 'perceive-reason-decide' process, an AI-Agent model for AMR is constructed, providing them with sensing, analyzing, and processing capabilities. Second, using the principles of single and bidirectional feedback modes, a collaborative interaction model is established to achieve their communication capabilities. Third, an AMR scheduling model is built based on Multi-agent Deep Deterministic Policy Gradient algorithm (MADDPG), which evaluates the scheduling result of transportation tasks by incorporating attention mechanisms and reward values. Finally, the collaborative scheduling scheme is validated through case studies. It can quickly respond to changes in material handling and make optimal allocation decisions, thus significantly improving the resilience of the workshop.
In modern product design, the modularisation of complex products plays a pivotal role by enabling greater design flexibility, reducing structural complexity, and improving overall manufacturability. Traditional methods of modularisation often rely heavily on expert knowledge and manual assessments, which can be time-consuming and error-prone. Despite increasing interest in automation, existing methods that integrate heterogeneous similarity and dependency metrics still suffer from limited automation and scalability, leaving room for more intelligent and efficient modularisation approaches. This paper proposes a new framework for the module division of complex products by integrating customised CAD tools, a Retrieval-Augmented Generation (RAG) pipeline supported by a Large Language Model (LLM), and a graph-based structural-attribute clustering algorithm. A case study on a bogie system demonstrates the feasibility of transforming dispersed design knowledge into quantifiable modular indicators, offering a practical solution for the automated modular design of complex products.
Aiming at the flexible job shop scheduling problem with limited AGV transportation, this paper presents an AGV and machine-integrated scheduling model with the optimisation objectives of minimising the maximum completion time and total energy consumption. It proposes an Improved SPEA2 Algorithm Combining DQN Algorithm (DQN-ISPEA2). The algorithm adopts a three-stage coding scheme for the workpiece, machine, and AGV. In the population initialisation stage, a hybrid initialisation strategy is employed to enhance the quality of the initial solution. DQN is used to adaptively select appropriate mutation operations, thereby enhancing the local search capability of the SPEA2 algorithm. Finally, it is compared with NSGA-III, NSGA-II, and SPEA2 algorithms on the MK and Kacem datasets. The experimental results demonstrate that the DQN-ISPEA2 algorithm yields satisfactory outcomes in most arithmetic cases, thereby verifying the feasibility and effectiveness of the proposed method in addressing the flexible job shop scheduling problem with limited AGV transportation. At the same time, the analysis of the impact of different AGV numbers on the optimisation objective reveals that the number of AGVs in the flexible workshop conforms to the law of diminishing marginal returns, providing a reference for actual manufacturing workshops when configuring AGVs.
Accurately identifying user needs is critical for producing emotional design. However, traditional user evaluation often suffers from randomness and subjective bias, which often lead to the loss of key information. To address this problem, this study proposes an approach that integrates interval-valued intuitionistic fuzzy sets (IVIFS) with the DEMATEL method to explore product form element relationships and identify key design elements, and using a CNN-SE-LSTM model to determine optimal combination of design solutions. First, users' emotional responses to products are systematically captured and categorised to reduce the dimensionality of target emotions. To handle uncertainty in expert evaluations and the directional relationships between product form elements, interval-valued intuitionistic fuzzy numbers (IVIFN) are employed, and IVIFS-DEMATEL method is then used to analyse the interaction relationships among smart cockpit form elements to extract core design elements. Second, the CNN-SE-LSTM model is applied to establish a nonlinear relationship between product form parameters and users' emotional needs, which could construct a predictive model for multidimensional perceptual requirements. Finally, an intelligent design system is developed to optimise product parameters based on consumers' perceptual needs. A case study of smart cockpit central control design demonstrates system effectiveness, with experimental results highlighting its potential to enhance customer satisfaction.
This study aims to explore the intricate relationship between technological affordance and users' affective agency within the context of medical live streaming, which is a kind of advanced digital service system. It seeks to understand how technology shapes user engagement and affective responses, which is crucial for enhancing user experience and the effectiveness of medical live streaming services. A novel analysis approach titled DHL approach is proposed (which combines the initial letters of each stage) that consists of three stages: (1) Data sourcing; (2) Human-AI topic modelling; (3) Logic formation. DHL, with a structural research framework incorporate with web text crawling, the LDA topic analysis, and Human-based topic regeneration and verification, and conceptual logic relationship formatting. The analysis distilled eight salient user-generated themes and mapped the sequential logic of technological affordance (digital affordance -> structural possibility -> service feasibility) alongside users' affective agency (motivation triggering -> emotional connection -> evaluative feedback). By translating these intertwined pathways into an explicit 'Technological affordance-Affective agency' relationship, the study delivers empirically grounded heuristics for service-design decisions. Thereby, abstract user signals could be converted into concrete, scenario-specific design specifications.
In the increasingly competitive New Energy Vehicle (NEV) market, consumers' purchase decisions are influenced not only by performance but also by emotional responses evoked by vehicle appearance. However, existing studies lack a robust framework to translate such preferences into concrete design parameters and to resolve the mapping between vague emotional needs and design features. To address this gap, this paper proposes an attractive NEV form design approach integrating the Interval Type-2 Trapezoidal Fuzzy Sets-Kano Model (IT2Tr-FKM) with the Hippopotamus Optimisation Algorithm-eXtreme Gradient Boosting (HO-XGBoost). Based on a three-level evaluation structure constructed using the Evaluation Grid Method (EGM) of Miryoku Engineering, IT2Tr-FKM is used to evaluate and prioritise upper-level Kansei words. Morphological decomposition then links key Kansei words to abstract reasons and concrete design attributes. HO-XGBoost establishes a mapping between Kansei words and representative design features to identify high emotional appeal solutions. Eye-tracking and the Data Information Difference Fluctuation Weighting Method (DIDF) are further applied for objective and subjective evaluation, ultimately selecting the optimal NEV form design. The proposed framework improves design efficiency and user satisfaction.
This paper investigates the scheduling of Circular Rail-Guided Vehicle (CRGV) in a real-world tobacco warehouse logistics system. In this system, CRGVs are required to transport materials among multiple material-handling units distributed across different areas. Since transportation are synchronised with material-handling operations, each task can only be transported after being released by the corresponding unit. Moreover, the diverse tasks among different units create a multi-type composite operation environment. These characteristics make efficient CRGV scheduling challenging, and fixed rule-based scheduling is therefore adopted in current industrial practice. To address the challenges, this study formulates a mathematical model with the objective of minimising the total flow time, which enables the processing of multi-type composite operations under task release time constraints. To improve solution quality, a heuristic-guided sequential small perturbation and a staged exponential cooling mechanism are proposed, resulting in an improved simulated annealing (ISA). Experiments based on real data demonstrate that, compared with the standard simulated annealing, ISA improves solution quality by 1.89% and obtains superior solutions relative to other methods. Furthermore, in scenarios with different task densities, the proposed approach achieves solutions approximately 10.75% better than existing rule-based scheduling, indicating strong robustness under varying task-density conditions and promising applicability in practical engineering.
In the product conceptual design process, the designers' ideation is highly dynamic and open-ended. Although existing generative AI expands the design space, it struggles to synchronise with the designer's cognitive state and perform deep collaborative reasoning. To address this, this paper proposes a real-time dynamic graph construction and reasoning method. This method adopts the Requirements-Functions-Logical-Physical (RFLP) ontology as a structural backbone, utilising large language models (LLMs) to extract concepts and relationships to construct structured logical chains. In the graph reasoning phase, the LLM performs path backtracking and node expansion based on the graph's topology to provide suggestions that drive the collaborative evolution of design ideas. We implemented this real-time visualised dynamic graph reasoning method within an interactive whiteboard prototype. A user study involving 24 participants reveals that, compared to a chat-only AI baseline, the proposed structured approach significantly improves the engineering quality of the final design schemes while ensuring the interpretability and controllability of the human-AI collaborative experience.
The transition to a circular economy calls for reduced reliance on fossil resources. Bio-based plastics offer potential environmental benefits, but their effective use in durable products is complex and under-researched. This study explores key considerations product developers face when using bio-based plastics in circular product development, with a focus on durable applications. Semi-structured interviews with product developers and a scoping literature review were conducted to identify and examine these considerations across the product life cycle. Eight key considerations were derived, highlighting dilemmas related to feedstock selection, regulations, material properties, the mass balance approach, costs, consumer perception, recovery strategies, and biodegradability. The study presents guidance to support product developers in navigating these considerations and making informed decisions. Results highlight the importance of early-stage life cycle thinking and interdisciplinary collaboration. Despite challenges, bio-based plastics can contribute to circular product development when supported by dedicated investments in knowledge and time to, for example, source bio-based plastics with low environmental impact and explore new design opportunities with novel bio-based plastics. The findings offer both theoretical insight and guidance for product developers aiming to incorporate bio-based plastics in their products.
Traditional research on product aesthetic evaluation has primarily relied on static images and outcome-oriented judgments, making it difficult to capture the formation process of aesthetic responses during dynamic viewing. To overcome this limitation, this study proposes a dynamic three-dimensional product aesthetic multimodal affective-cognitive integration (MACI) framework. Building on this framework, we propose a MACI-based guided design evaluation method tailored for dynamic product evaluation. This method organises multimodal physiological signals into process-based evidence, linking attention allocation, affective arousal, and cognitive processing to design-related evaluation outcomes. The study conducted a foundational validation study using dynamic 3D automotive presentation videos combined with eye-tracking, electrodermal activity (EDA), and electroencephalogram (EEG) data. The results indicate that the proposed method can reliably distinguish between different aesthetic levels; the KNN-based implementation achieved a test accuracy of 95.45% and an average accuracy of 92.88% in a 10-fold cross-validation. More importantly, multimodal evidence was further translated into interpretable design cues related to visual organisation, cognitive integration, and iterative refinement. Consequently, this study not only provides an empirical validation case but also establishes a process-oriented methodological pathway for integrating multimodal evidence into design review and optimisation processes.
The proper application of evaluation tools is critical to ensuring the usability of human-machine interaction (HMI) in intelligent vehicles. However, a systematic understanding of the current usage patterns, applicability boundaries, and methodological issues of evaluation tools remains absent in this field. To help practitioners conduct evaluations effectively, this study systematically analyses the evaluation tools used across 105 applied research cases in this field. The results indicate that the field has established methodological paradigms of balancing preference report tools with performance observation tools, delineating the roles of user testing and expert inspection tools, and implementing multi-method and mixed-method evaluation. Nevertheless, practitioners still exhibit a tendency toward excessive reliance on subjective preference tools in practice, expert inspection tools such as heuristic evaluation and human performance modeling (HPM) have long been absent, general HCI questionnaire scales are adopted without considering their adaptability to driving scenarios, and the cross-validation function of mixed-method evaluation is largely overlooked. To address these challenges, this study proposes actionable recommendations and distils decision guidelines at both theoretical and practical levels, thereby providing practitioners with concrete guidance for transitioning from general-purpose screening to evidence-based matching and advancing evaluation practice in this field from experience-driven approaches toward scientific standardisation.
Current generative design schemes for product styling are primarily limited to two-dimensional space, which fails to adequately represent the 3D volumetric attributes and hinders the direct translation of emotional intent into manufacturable forms. To overcome this limitation, this study proposes the ENM (Emotion data-NST-MVSNet) framework, a unified pipeline that integrates deep learning classification, style transfer, and multi-view stereo reconstruction. First, ResNet18 is employed to construct a large-scale emotional dataset, quantifying the implicit mapping between user emotional requirements and visual styling features. Second, Neural Style Transfer (NST) is utilised to generate emotionally compliant multi-view images while preserving semantic consistency. Third, an enhanced MVSNet, incorporating deformable convolutions and a distance-aware loss re-weighting strategy, is developed to robustly reconstruct these stylised images into high-fidelity 3D models, enabling a seamless transition from 2D emotional cues to 3D geometric forms. Validated through an automotive styling case study, the experimental results demonstrate that the proposed method achieves high reconstruction quality and aesthetic satisfaction, with user evaluation scores exceeding 4.3 across key emotional dimensions. This research provides a high-efficiency pathway for intelligent product styling design and lays a solid foundation for automated, emotion-driven 3D product prototyping and future cross-dimensional generative design.
