This study proposes an integrated Interval Type-2 Fuzzy Analytic Network Process-VIseKriterijumska Optimizacija I Kompromisno Resenje for evaluating digital green mechanical engineering curricula, addressing the complexity of sustainability-focused education. The method combines interval type-2 fuzzy sets to handle linguistic uncertainties in expert judgments, Analytic Network process to model interdependencies among evaluation criteria, and VIKOR to generate dynamic compromise solutions for conflicting objectives. A real-world case study demonstrates the framework’s applicability, identifying the optimal curriculum design based on technical, environmental, and pedagogical criteria. Sensitivity and comparative analyses confirm the method’s robustness, which outperforms conventional approaches in managing higher-order uncertainties. The study contributes a reliable decision-making tool for aligning mechanical engineering education with sustainability goals while balancing academic rigor and industrial relevance.

Human-robot collaborative partial disassembly lines offer a promising approach for end-of-life product recovery, and their balancing has received growing attention. However, existing studies still have several limitations. They often rely on deterministic disassembly information or stochastic formulations requiring hard-to-obtain complete probability distributions, pay limited attention to interactive collaboration among heterogeneous human and robot resources, and ignore degradation-related failure risks. To address these gaps, this paper formulates a heterogeneous interactive human-robot collaborative partial disassembly line balancing problem with preventive maintenance under type-2 fuzzy uncertainty. The proposed formulation incorporates interactive collaboration among heterogeneous human and robot resources, introduces preventive maintenance to address degradation-related failure risks, and models uncertainty using interval type-2 triangular fuzzy numbers. The objective is to jointly minimise the regular scenario cycle time, the total cycle time across preventive maintenance scenarios, and the reconfiguration changes induced by maintenance. To solve this problem, a Q-learning-guided bees algorithm is developed. The algorithm adopts a problem-oriented four-layer encoding scheme and a heuristic decoding mechanism, integrates three neighbourhood operators and three enhanced search operators, and uses Q-learning to adaptively select operators to balance exploration and exploitation. The proposed model and algorithm are validated using a transmission disassembly case, confirming the benefits of partial disassembly, heterogeneous human and robot resources, interactive collaboration, preventive maintenance, and type-2 fuzzy uncertainty. A sensitivity analysis is then conducted to quantify the impact of uncertainty levels on model outputs. Finally, comparisons with CPLEX and three advanced metaheuristics demonstrate the competitiveness of the proposed algorithm.

This paper introduces a decentralised task planning and motion coordination (TPMC) framework for scalable multi-robot collaborative manufacturing (MRCMfg). Built upon cognitive digital twins, the framework addresses challenges in dynamic system configurations, such as real-time adjustments to the number and position of active robots. The manufacturing process is divided into four phases: cognition, configuration, planning, and coordination. A large language model (LLM) analyses and infers data from cognitive digital twin simulations for state cognition and configuration optimisation, ensuring scalability by dynamically adjusting robot layouts. Multi-agent deep reinforcement learning (MADRL) is employed for TPMC, which relies solely on local observations and incorporates a Transformer, along with a Soft Mixture-of-Experts (MoE) module, to update the policy network. This results in a decentralised decision-making policy with superior scalability and generalisation capabilities. A novel decentralised Soft-Soft Actor-Critic (SSAC) algorithm is developed, integrating a Soft MoE into the policy network, enhancing policy generalisation. The method is trained and tested on a scalable MRCMfg system in cognitive digital twins, specifically for the disassembly of electric vehicle batteries. Experiments demonstrate the approach's effectiveness, scalability, and efficiency in handling dynamic system configurations, as well as achieving efficient TPMC. The paper concludes by highlighting the framework's ability to maintain continuous operation in dynamic environments. The paper also suggests future work on enhancing the coupling mechanisms between task and motion planning (TAMP), integrating stochastic task-duration modelling, and incorporating physical priors into training.

Disassembly lines dismantle scrapped products to retain valuable components for recycling and remanufacturing. Many factors affect the productivity of disassembly lines, especially the operating cost of the workstation, the priority of the disassembly task, the skill level of the workers, and the learning proficiency. This study considers the learning effect of disassembly and assembly workers. It optimizes two objectives simultaneously: 1) maximizing the disassembly profit and 2) maximizing the learning outcomes of the workers. A bi-objective mixed integer programming model for the disassembly and assembly balance problem is established to explore the optimal solution. A multi-objective fruit fly optimization algorithm is proposed, which can better solve multi-objective optimization problems by simulating the foraging behaviour of fruit flies and finding the optimal solution set of multiple optimization objectives. The algorithm is compared with the CPLEX solver and other multi-objective optimization algorithms. The experimental results show that the algorithm has obvious advantages.

During human-robot collaboration (HRC), robots share workplaces with humans, and there may be frequent contact between them. It is crucial to be able to detect unexpected collisions in real-time so that appropriate safety measures should be taken to avoid injuries to humans and damage to robots. However, there are challenges with existing collision detection strategies, such as the additional costs incurred in deploying sensors in robots to implement pre-collision safety surveillance solutions or conducting complicated experiments to develop post-collision compliance solutions. To address these challenges, this paper presents a new momentum observerbased collision detection approach in which the external torques caused by collisions on robots can be efficiently identified. The approach involves integrating an improved deep Lagrangian network (DeLaN) to model robot dynamics without dynamic parameter identification experiments and prior knowledge of the robot's physical and structural parameters. Another innovation of this approach is that a compensatory safety threshold is designed to enhance collision detection accuracy. Three robot datasets were used to train the improved DeLaN model. Simulation and real-world experiments were further carried out on the proposed approach to validate the effectiveness of the approach. Comparative experiments showed that the proposed approach outperformed other momentum observers in terms of both speed and efficiency. Moreover, experiments showed that the compensatory safety threshold proposed in this approach mitigated false positives caused by friction errors in robot joints to prevent the misdetection of collisions.

Remanufacturing scheduling is a critical path to sustainability. However, existing research on the three-stage remanufacturing system scheduling problem (3T-RSSP) is predominantly concentrated in the deterministic domain; the few studies involving uncertainty mainly employ stochastic programming or symmetric fuzzy sets, neglecting the asymmetric uncertainty inherent in remanufacturing. This paper investigates the bi-objective energy-aware 3T-RSSP under asymmetric uncertainty, integrating disassembly, reprocessing, and reassembly. First, a fuzzy mathematical model is formulated to minimize makespan and energy consumption; by quantifying aggregate asymmetric uncertainty via a Right-Skewed Perturbation Model based on Triangular Fuzzy Numbers, it captures operational delay characteristics to accurately map real-world scenarios. Second, an Uncertainty-Aware Scheduling Strategy (UASS) is proposed to preserve the stability of the scheduling scheme. By treating tasks with similar expected processing times as indistinguishable ones and using the Coefficient of Variation as a tie-breaking criterion, UASS implements a risk postponement mechanism to shift high-volatility operations to later stages. Third, an Improved Hybrid Guided Whale Migration Algorithm (HGWMA) is proposed, incorporating AGALF and PAVM strategies to balance exploration and exploitation. Finally, comprehensive simulations across varying scales evaluate HGWMA against other algorithms. Compared with NSGA-II, HGWMA achieves a 12.8% average Hypervolume improvement and significantly reduces Inverted Generational Distance, especially in large-scale instances. Results demonstrate the feasibility and superiority of the proposed framework in addressing the energy-aware scheduling problem under asymmetric uncertainty.

To advance automation in the remanufacturing sector, this paper proposes a novel spatial disassembly constraint modelling method that integrates cross-modal deep learning with the reasoning capabilities of Large Language Models (LLMs). A significant challenge in robotic disassembly is accurately identifying spatial constraints for used products, which often exhibit uncertain conditions and structural deviations from original designs. To address this, we introduce a cross-modal deep learning technique that unifies the detection mechanism for both captured images of used products and CAD models of original products, effectively resolving multimodal data fusion issues during model training. Furthermore, we leverage LLMs to construct and match semantic graphs based on the spatial positional relationships among components. This allows for the precise identification of the optimal matching CAD model and the correct numbering of detected components, thereby deriving specific spatial constraints for each component. The rigour and performance of this method are validated through a case study involving the removal of used bolster springs from railway wagon bogies on a robotic platform. The results demonstrate that the proposed method enables robots to accurately construct spatial disassembly constraints under complex and uncertain conditions, facilitating automated disassembly operations.

Research on scheduling of three-stage remanufacturing systems has received increasing attention. However, most existing studies adopt parallel disassembly/assembly configurations, which can increase makespan and energy consumption, motivating the development of a more suitable system layout and scheduling model. To address this gap, this study proposes an energy-aware three-stage heterogeneous flexible-job-type remanufacturing scheduling problem (3T-HFRSSP) based on a novel configuration integrating type-II disassembly/assembly lines with a flexible-job-type reprocessing shop. An energy-aware bi-objective mathematical model is formulated to minimize makespan and total energy consumption. To solve the proposed problem, an enhanced Jaya-NSGA algorithm with a five-layer encoding scheme is developed by integrating the directional search mechanism of the Jaya algorithm with the non-dominated sorting, crossover, and mutation mechanisms of the NSGA II algorithm. Experiments on four product-scale instances and comparative studies on eleven instances demonstrate the effectiveness of the proposed model and algorithm. Results show that the proposed 3T-HFRSSP significantly reduces makespan compared with traditional three-stage remanufacturing configurations, and its advantage becomes more pronounced as the EOL product scale and the number of disassembly workstations increase. The proposed Jaya-NSGA algorithm also shows strong competitiveness in solving the 3T-HFRSSP. This study provides a useful perspective for green remanufacturing and scheduling optimization of three-stage remanufacturing systems.

Remanufacturing is a key strategy for implementing closed-loop supply chains, but its adoption faces challenges in controlling the end-of-life product reverse flows and achieving customer acceptance due to uncertain perceived quality. Leasing of products can address these issues, facilitating remanufacturing. Indeed, leasing allows the manufacturer to keep product ownership, facilitating the return of end-of-life products. Similarly, leasing alleviates customers from the burden of maintenance, potentially leading to higher customer acceptance. The integration of remanufacturing and leasing, referred to as “remanu-leasing”, involves three main stakeholders: the manufacturer, the customer, and the environment. While remanu-leasing can potentially align the interests of the manufacturer and customer, the overall benefit to all stakeholders is often overlooked. This paper investigates when remanu-leasing provides mutual benefits for all three parties. Mathematical models are developed to represent the objectives of each stakeholder. Using these models, various scenarios are generated to explore a wide range of outcomes. Key parameters influencing remanu-leasing success are identified, and decision trees are created for quick assessment of its viability. The results demonstrate that remanu-leasing benefits all stakeholders under specific conditions, even though they can be limited. The study reveals that remanu-leasing tends to be advantageous for the manufacturer and the environment when remanufacturing volumes and manufacturing preservation rates are low. For customers, remanu-leasing is preferred when the mean time to failure and leasing fees are low. These findings provide valuable insights for managers seeking to implement remanu-leasing, highlighting the key parameters to consider for enhancing its feasibility.

Robot vision has been proposed for detection and tracking tasks in autonomous disassembly. At the same time, the high uncertainty of end-of-life (EoL) products can affect the performance of robot vision systems. In this paper, we propose an EoL object detection model that allows accurate and robust detection based on the training data of new products. This technique is helpful in vision-based robotic disassembly, where the data of old or EoL products may not be available for training. We first develop a dual-constraint region proposal mechanism, in which the relationships between anchors and target regions, corners and target regions, and anchors and corners are integrated to describe EoL products based on the pair of anchors and corners. The region proposal mechanism is further introduced to the design of the EoL object detection model, where a three-level feature fusion module is developed to fuse and align the adjacent level of multi-scale anchor features and corner features that are extracted by the existing backbone module. A localisation loss function is designed to optimise the proposed model by maximising the interaction-over-union between ground-truth regions and predicted regions represented by anchors, corners, and the pair of anchors and corners, respectively. The performance of the proposed EoL object detection model is validated based on a customised dataset, demonstrating the contribution of this research to eliminate the limitation of the uncertainty of EoL products on vision-based robotic disassembly. With the proposed model, the robot vision system yields accurate and robust EoL object detection in the robotic disassembly of EoL electric vehicle batteries, demonstrated by a screw-removal case.

This paper presents two novel harmonically disturbance-resistant zeroing neural network (ZNN) models: the known frequency harmonic-resistant ZNN (KFHRZNN) and the unknown frequency harmonic-resistant ZNN (UFHRZNN). These models are designed to tackle the pseudoinverse of time-varying matrices and inverse kinematics challenges in robotic manipulators. By precisely accounting for the derivatives of harmonic disturbances, they significantly mitigate these interferences, thereby improving the control efficacy of robots in high-speed, dynamic settings. The study elucidates the design rationale, convergence characteristics, and stability assessments for both KFHRZNN and UFHRZNN. Numerical simulations and physical experiments validate the effectiveness and advantages of these models in resolving time-varying issues within robotic manipulators, highlighting their precision and robustness against harmonic disturbances.

Collaborative human-robot units have attracted recognition for their ability to be used for flexible product disassembly to help achieve intelligent, sustainable, and service-oriented remanufacturing. The adoption of human-robot collaborative disassembly (HRCD) in Industry 5.0 contributes to enhancing the flexibility of the eco-friendly sustainable manufacturing supply chain, realising a circular product life cycle, and facilitating the transition to carbon neutrality. To conduct a systematic examination of the development and research trends in HRCD, a quantitative analysis was carried out on 99 studies retrieved from databases. The research topic structure was examined from an array of perspectives, shedding light on the current state, future pathways, and focal areas in conjunction with a visually depicted knowledge graph. This paper enables scholars to comprehend the trajectory and pivotal aspects of HRCD through investigations of intelligent remanufacturing, thereby clarifying the path for further advancements in sustainable manufacturing practices.

In recent years, the rapid depletion of natural resources and irreversible environmental damage have underscored the significance of a Circular Economy. The Circular Economy aims to minimise waste, optimise resource utilisation, and extend product lifespans. However, while the direct recycling of resources produces some positive outcomes, the greenhouse gases emitted during this process remain a substantial threat to the environment. Consequently, remanufacturing has emerged as a favoured option. Disassembly sequence planning (DSP) is a critical initial step in remanufacturing. DSP is classified as an NP-Hard problem, indicating that its solution is computationally complex, rendering traditional methods impractical, particularly for large-scale problems. To address these challenges, metaheuristic optimisation algorithms inspired by natural processes offer effective and efficient solutions. This paper proposes a novel method that integrates the Bees Algorithm (BA) with reinforcement learning (RL) techniques to solve the DSP problem, with a focus on minimising disassembly time. The Bees Algorithm, introduced in 2005 to emulate the foraging behaviour of honeybees, has seen numerous variations. The objective of incorporating RL is to enhance the adaptability of the Bees Algorithm to the search space. To validate the efficiency of the proposed method, the results were compared with the existing BA-based solutions in the literature. This comparative analysis aims to demonstrate the superiority and practical applicability of the new approach in addressing the complexities of DSP.

This paper introduces the Beetle Olfactory-based Manipulability Optimizer Recurrent Neural Network (BOMO-RNN), an advanced RNN-based controller designed to enhance the manipulability of redundantly actuated industrial robotic arms. The manipulability index, which quantifies the maneuverability of the robotic arm, is crucial for avoiding kinematic singularities that restrict the mobility of robotic arm in the task space. The proposed approach formulates an optimisation problem using the penalty method to incorporate the manipulability index into the tracking control objective function. Unlike conventional approaches that rely on velocity-level control and require precise initialisation, BOMO-RNN operates at the position level, allowing direct trajectory tracking from arbitrary starting configurations, thereby increasing flexibility and ease of deployment. This function aims to maximise maneuverability while ensuring accurate tracking of the reference trajectory, effectively avoiding joint-space singularities. The BOMO-RNN framework leverages a metaheuristic optimisation strategy, enabling efficient exploration of high-dimensional search spaces without requiring explicit Jacobian pseudo-inversion, significantly reducing computational overhead and improving numerical stability. The BOMO-RNN algorithm efficiently addresses the time-varying optimisation problem at the position level, eliminating the need for computationally intensive Jacobian pseudo-inversion. This ensures robustness in real-world scenarios where high-speed control and adaptability to dynamic environments are critical. The algorithm's convergence is theoretically analysed, and its performance is validated through numerical simulations and experimental results using the LBR IIWA 7-DOF robot. Extensive experimental verification demonstrates the effectiveness of BOMO-RNN across diverse trajectory patterns, including circular, sinusoidal, and piecewise straight-line motions, confirming its generalizability and practical applicability. The results demonstrate BOMO-RNN's practical effectiveness in optimising manipulability and its potential for real-world robotic applications.

Enhancing the recycling efficiency of end-of-life (EOL) products is crucial for promoting a circular economy. Disassembly sequence planning (DSP) is a key technology in this process. However, traditional DSP relies on manually executed sequential operations, resulting in inefficiency and high labor intensity, while fully automated robotic disassembly, without human involvement, struggles to independently complete tasks. To address this, human-robot collaborative disassembly sequence planning (HRCDSP) has emerged, where humans and robots perform disassembly tasks in parallel to enhance efficiency and reduce labor demands. Despite its potential, existing HRCDSP studies often assume deterministic disassembly conditions, overlook real-world uncertainties, and favor complete disassembly, which is often unnecessary in practice. Furthermore, current research has yet to systematically balance the trade-offs among disassembly time, energy consumption, and profit, thereby limiting practical applicability. To address these gaps, this paper proposes a chance-constrained programming-based human-robot collaborative selective disassembly sequence planning (CHRCSDSP) problem. This approach integrates uncertainties and selective disassembly into HRCDSP decision-making, enabling decision-makers to flexibly balance disassembly time, energy consumption, and profit under predefined confidence levels. Considering the complexity of this problem, an enhanced multi-objective bees algorithm (EMOBA) is proposed, integrating reinforcement learning and variable neighborhood search (VNS) strategies to improve solution performance. Experimental results demonstrate that CHRCSDSP outperforms human-only and robot-only disassembly in feasibility and effectiveness, while EMOBA surpasses other advanced algorithms. This study strengthens HRCDSP's real-world applicability, contributing to more efficient EOL product recycling.

Remanufacturing has become a mainstream sustainable manufacturing paradigm for energy conservation and environmental protection. Disassembly and reprocessing operations are two main activities in remanufacturing. This work proposes multiobjective integrated scheduling of disassembly and reprocessing operations considering product structures and random processing time. First, a stochastic programming model is developed to minimize maximum completion time and total tardiness. Second, a reinforcement learning-based multiobjective evolutionary algorithm is devised considering problem-specific knowledge. Three search strategy combinations are formed: crossover and mutation, crossover and key product-based iterated local search, mutation and key product-based iterated local search. At each iteration, a Q-learning method is devised to intelligently choose a combination of premium strategies. A stochastic simulation is incorporated to evaluate the objective values of the searched solutions. Finally, the formulated model and method are compared with an exact solver, CPLEX, and three well-known metaheuristics from the literature on a set of test instances. The results confirm the excellent competitiveness of the developed model and algorithm for solving the considered problem.