While Augmented Reality (AR) offers the potential to provide real-time guidance, one of the barriers to its adoption in industrial assembly is the lack of fast, no-code, intelligent methods for generating AR-assisted assembly programs. This paper proposes an AI-aided AR-Assisted Assembly Instruction Authoring and Generation method (ARAIAG) to address these challenges. ARAIAG allows engineers, without coding expertise, to intuitively design AR-assisted assembly instructions based on assembly demonstrations captured through RGBD cameras. Based on ARAIAG, we propose a new algorithm considering hand manipulation and model characteristics to achieve spatial registration for models, virtual-physical fusion, and assembly direction recognition. Additionally, we employed a novel human–computer interaction method and Large Language Model (LLM)-assisted content generation to achieve the automatic creation of interactive and instructive AR-assisted assembly programs. Through this approach, we streamline program development and enable more efficient AR-assisted assembly in dynamic manufacturing environments.
{"title":"AI-aided Automated AR-Assisted Assembly Instruction Authoring and Generation method","authors":"Junjian Lin, Jianjian Wang, Pingfa Feng, Xiangyu Zhang, Dingwen Yu, Jianfu Zhang","doi":"10.1016/j.jmsy.2025.08.019","DOIUrl":"10.1016/j.jmsy.2025.08.019","url":null,"abstract":"<div><div>While Augmented Reality (AR) offers the potential to provide real-time guidance, one of the barriers to its adoption in industrial assembly is the lack of fast, no-code, intelligent methods for generating AR-assisted assembly programs. This paper proposes an AI-aided AR-Assisted Assembly Instruction Authoring and Generation method (ARAIAG) to address these challenges. ARAIAG allows engineers, without coding expertise, to intuitively design AR-assisted assembly instructions based on assembly demonstrations captured through RGBD cameras. Based on ARAIAG, we propose a new algorithm considering hand manipulation and model characteristics to achieve spatial registration for models, virtual-physical fusion, and assembly direction recognition. Additionally, we employed a novel human–computer interaction method and Large Language Model (LLM)-assisted content generation to achieve the automatic creation of interactive and instructive AR-assisted assembly programs. Through this approach, we streamline program development and enable more efficient AR-assisted assembly in dynamic manufacturing environments.</div></div>","PeriodicalId":16227,"journal":{"name":"Journal of Manufacturing Systems","volume":"83 ","pages":"Pages 405-423"},"PeriodicalIF":14.2,"publicationDate":"2025-09-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145220750","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-09-26DOI: 10.1016/j.jmsy.2025.09.015
Yulu Zhou , Shichang Du , Jun Lv , Xiaoxiao Shen , Andrea Matta , Siyang Wang
Pallet automation system (PAS) is crucial for enterprises to organize and schedule limited resources, such as fixture-pallets (FPs) and material-pallets (MPs). In customized production, FPs are often insufficient and unbalanced. To address this, MPs are prepared to store workpieces to release FPs' capacity. In this way, FPs are utilized for processing, while MPs are leveraged for storage. However, existing studies mainly focus on fixtures that are fixed to machines and rarely consider FPs and MPs. To address this gap, this paper investigates the flexible pallet automation system scheduling with limited FPs and MPs (FPASFM). Firstly, a mathematical model is established to minimize the makespan. Secondly, a five-layer encoding strategy, a new decoding method, and a feasibility correction strategy are integrated to obtain feasible solutions. Thirdly, an improved meta-heuristic algorithm with rule-based initialization and critical path mutation (IMHRC) is proposed. Finally, effective initialization rule combinations are identified through experiments with 36 different rule combinations. 15 real-data case studies show that IMHRC outperforms six other algorithms. Additionally, IMHRC significantly reduces makespan by 59.66 % and 45.90 % for two real orders, while enhancing resource utilization. IMHRC demonstrates the ability to obtain superior solutions in a shorter time, with its advantages in large-scale problems, effectively meeting the practical demands of enterprises in real-world production environments.
{"title":"Flexible pallet automation system scheduling with limited fixture-pallets and material-pallets: A case study from an engine manufacturing enterprise","authors":"Yulu Zhou , Shichang Du , Jun Lv , Xiaoxiao Shen , Andrea Matta , Siyang Wang","doi":"10.1016/j.jmsy.2025.09.015","DOIUrl":"10.1016/j.jmsy.2025.09.015","url":null,"abstract":"<div><div>Pallet automation system (PAS) is crucial for enterprises to organize and schedule limited resources, such as fixture-pallets (FPs) and material-pallets (MPs). In customized production, FPs are often insufficient and unbalanced. To address this, MPs are prepared to store workpieces to release FPs' capacity. In this way, FPs are utilized for processing, while MPs are leveraged for storage. However, existing studies mainly focus on fixtures that are fixed to machines and rarely consider FPs and MPs. To address this gap, this paper investigates the flexible pallet automation system scheduling with limited FPs and MPs (FPASFM). Firstly, a mathematical model is established to minimize the makespan. Secondly, a five-layer encoding strategy, a new decoding method, and a feasibility correction strategy are integrated to obtain feasible solutions. Thirdly, an improved meta-heuristic algorithm with rule-based initialization and critical path mutation (IMHRC) is proposed. Finally, effective initialization rule combinations are identified through experiments with 36 different rule combinations. 15 real-data case studies show that IMHRC outperforms six other algorithms. Additionally, IMHRC significantly reduces makespan by 59.66 % and 45.90 % for two real orders, while enhancing resource utilization. IMHRC demonstrates the ability to obtain superior solutions in a shorter time, with its advantages in large-scale problems, effectively meeting the practical demands of enterprises in real-world production environments.</div></div>","PeriodicalId":16227,"journal":{"name":"Journal of Manufacturing Systems","volume":"83 ","pages":"Pages 357-371"},"PeriodicalIF":14.2,"publicationDate":"2025-09-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145155325","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-09-26DOI: 10.1016/j.jmsy.2025.09.013
Ruihao Kang , Junshan Hu , Xingtao Su , Zhengping Li , Zhanghu Shi , Wei Tian
Digital Twin (DT) technology is one of the key approaches to enhancing the intelligence of aircraft assembly equipment. However, the diversity of such equipment types and significant structural differences present substantial challenges to the development of DT models. This article proposes a unified V-shaped DT modeling paradigm to support high-accuracy and structured modeling. The robotic drilling system is used as an example to validate this paradigm. The modeling requirements of this system are established based on a comprehensive analysis of its structural characteristics and operational tasks. A corresponding virtual entity is constructed through parametric modeling based on kinematic analysis. The behavior model represents the interaction protocols and decision logic of the physical system, with basic modules for communication and behavioral analysis. These modules are then systematically integrated to form a complete task model for drilling. The structural validation of the virtual entity is performed, accompanied by the formulation of behavioral matching degree and task execution consistency to evaluate the effectiveness of the proposed modeling paradigm. Meanwhile, kinematic parameter identification is integrated to calibrate the virtual entity, thereby further enhancing the DT modeling accuracy. The experimental results show that the behavior matching degree for positioning after calibration is 0.204 ± 0.228 mm, with an increase of 78.71 %. The average errors of hole position and diameter are reduced by 78.43 % and 14.27 %, respectively, after calibration. The corresponding task execution consistency is improved to 1.465 and 1.462. This indicates that the high-accuracy DT model constructed by the proposed paradigm effectively enhances the intelligence and assembly quality of the equipment.
{"title":"A unified V-shaped digital twin modeling paradigm of aircraft assembly systems for improving modeling accuracy and assembly quality","authors":"Ruihao Kang , Junshan Hu , Xingtao Su , Zhengping Li , Zhanghu Shi , Wei Tian","doi":"10.1016/j.jmsy.2025.09.013","DOIUrl":"10.1016/j.jmsy.2025.09.013","url":null,"abstract":"<div><div>Digital Twin (DT) technology is one of the key approaches to enhancing the intelligence of aircraft assembly equipment. However, the diversity of such equipment types and significant structural differences present substantial challenges to the development of DT models. This article proposes a unified V-shaped DT modeling paradigm to support high-accuracy and structured modeling. The robotic drilling system is used as an example to validate this paradigm. The modeling requirements of this system are established based on a comprehensive analysis of its structural characteristics and operational tasks. A corresponding virtual entity is constructed through parametric modeling based on kinematic analysis. The behavior model represents the interaction protocols and decision logic of the physical system, with basic modules for communication and behavioral analysis. These modules are then systematically integrated to form a complete task model for drilling. The structural validation of the virtual entity is performed, accompanied by the formulation of behavioral matching degree and task execution consistency to evaluate the effectiveness of the proposed modeling paradigm. Meanwhile, kinematic parameter identification is integrated to calibrate the virtual entity, thereby further enhancing the DT modeling accuracy. The experimental results show that the behavior matching degree for positioning after calibration is 0.204 ± 0.228 mm, with an increase of 78.71 %. The average errors of hole position and diameter are reduced by 78.43 % and 14.27 %, respectively, after calibration. The corresponding task execution consistency is improved to 1.465 and 1.462. This indicates that the high-accuracy DT model constructed by the proposed paradigm effectively enhances the intelligence and assembly quality of the equipment.</div></div>","PeriodicalId":16227,"journal":{"name":"Journal of Manufacturing Systems","volume":"83 ","pages":"Pages 372-391"},"PeriodicalIF":14.2,"publicationDate":"2025-09-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145155081","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-09-24DOI: 10.1016/j.jmsy.2025.09.012
Kyu-Tae Park , Chiho Lim , Ju-Yong Lee
Human–robot collaboration (HRC) is a key enabler of human-centric manufacturing, achieved through cooperation between human operators and collaborative robots. HRC can be classified into three developmental phases: coexistence, sequential collaboration, and simultaneous cooperation. To address ergonomic fatigue and simultaneous cooperation (HRCAW-ES) constraints, this study introduces a novel scheduling model that integrates sequential collaboration and simultaneous cooperation, focusing on production scheduling in shared HRC assembly workstations involving one human operator and one collaborative robot. This setting accounts for key operational constraints, including operation precedence and assembly relationships, human task eligibility based on ergonomic risk factors, ergonomic fatigue accumulation and recovery following established models, sequence-dependent setup for end-effector switching on a collaborative robot, and simultaneous cooperation between the two collaborators. A mathematical model was developed to formulate an adaptive variable neighbourhood search (AVNS) algorithm and a disjunctive graph representation was employed to analyse the structural characteristics of the HRCAW-ES. An ablation study performed using both linear and nonlinear fatigue models revealed the superior performance of the proposed AVNS algorithm compared to the control group across various scenarios involving varying cooperation ratio and fatigue levels. This experiment includes results obtained using parameters collected from the small-product packaging and cable-assembly processes. Emphasis was placed on examining the impacts of ergonomic limitations and simultaneous cooperation within the scheduling framework. The proposed method generates high-quality, feasible schedules to address the complexity introduced by ergonomic constraints and cooperative requirements. The method may be extendable to a wide range of assembling processes where full automation is infeasible.
{"title":"Production scheduling for human–robot collaborative assembly workstations under constraints of ergonomic fatigue and simultaneous cooperation","authors":"Kyu-Tae Park , Chiho Lim , Ju-Yong Lee","doi":"10.1016/j.jmsy.2025.09.012","DOIUrl":"10.1016/j.jmsy.2025.09.012","url":null,"abstract":"<div><div>Human–robot collaboration (HRC) is a key enabler of human-centric manufacturing, achieved through cooperation between human operators and collaborative robots. HRC can be classified into three developmental phases: coexistence, sequential collaboration, and simultaneous cooperation. To address ergonomic fatigue and simultaneous cooperation (HRCAW-ES) constraints, this study introduces a novel scheduling model that integrates sequential collaboration and simultaneous cooperation, focusing on production scheduling in shared HRC assembly workstations involving one human operator and one collaborative robot. This setting accounts for key operational constraints, including operation precedence and assembly relationships, human task eligibility based on ergonomic risk factors, ergonomic fatigue accumulation and recovery following established models, sequence-dependent setup for end-effector switching on a collaborative robot, and simultaneous cooperation between the two collaborators. A mathematical model was developed to formulate an adaptive variable neighbourhood search (AVNS) algorithm and a disjunctive graph representation was employed to analyse the structural characteristics of the HRCAW-ES. An ablation study performed using both linear and nonlinear fatigue models revealed the superior performance of the proposed AVNS algorithm compared to the control group across various scenarios involving varying cooperation ratio and fatigue levels. This experiment includes results obtained using parameters collected from the small-product packaging and cable-assembly processes. Emphasis was placed on examining the impacts of ergonomic limitations and simultaneous cooperation within the scheduling framework. The proposed method generates high-quality, feasible schedules to address the complexity introduced by ergonomic constraints and cooperative requirements. The method may be extendable to a wide range of assembling processes where full automation is infeasible.</div></div>","PeriodicalId":16227,"journal":{"name":"Journal of Manufacturing Systems","volume":"83 ","pages":"Pages 337-356"},"PeriodicalIF":14.2,"publicationDate":"2025-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145155082","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-09-23DOI: 10.1016/j.jmsy.2025.09.010
Liu Xinyu , Yuan Bingkun , Wang Pengchao , Ding Ning , Chu Jianjie
Recent advances in Large Language Models (LLMs) have demonstrated their unparalleled capability in collaborative design requirements mining, offering significant potential for the integration of Human–Machine Teams (HMTs) and improved efficiency in design mining processes. However, existing approaches often lack integrated frameworks capable of simultaneously addressing both the compositional and operational challenges of LLM–human teams, which hinders their effective deployment in complex, real-world scenarios. Specifically, two critical challenges remain: first, how to effectively transform LLMs into reliable domain experts capable of understanding and elaborating design requirements; and second, how to optimize HMT configurations amid inherent ambiguities in human expert evaluations. To address these gaps, we propose a novel dual-paradigm Hybrid-Augmented Intelligence (HAI) framework that integrates Cognitive Computing (CC-HAI) with Human-in-the-Loop (HITL-HAI) mechanisms. Our key contributions include a CC-HAI–based cognitive teammate mechanism that uses structured prompt engineering to transform LLMs into domain-specialized roles, facilitating the formation of collaborative HMTs; and an HITL-HAI uncertainty mitigation method that employs a Z-number-enhanced cloud modeling approach to manage subjective uncertainties in expert assessments and support robust team configuration. The framework is validated through multi-domain case studies spanning smart home systems, smart cockpits, medical devices, and baby products. Extensive experiments demonstrate its effectiveness in terms of team performance, error reduction, cross-domain generalizability, and decision-making superiority. This research provides a replicable paradigm for deploying LLMs as cognitive collaborators in collaborative design ecosystems, contributing to both theory and methodology in human–machine team intelligence.
{"title":"An approach based on hybrid-augmented intelligence for the combination and optimization of human-machine teams","authors":"Liu Xinyu , Yuan Bingkun , Wang Pengchao , Ding Ning , Chu Jianjie","doi":"10.1016/j.jmsy.2025.09.010","DOIUrl":"10.1016/j.jmsy.2025.09.010","url":null,"abstract":"<div><div>Recent advances in Large Language Models (LLMs) have demonstrated their unparalleled capability in collaborative design requirements mining, offering significant potential for the integration of Human–Machine Teams (HMTs) and improved efficiency in design mining processes. However, existing approaches often lack integrated frameworks capable of simultaneously addressing both the compositional and operational challenges of LLM–human teams, which hinders their effective deployment in complex, real-world scenarios. Specifically, two critical challenges remain: first, how to effectively transform LLMs into reliable domain experts capable of understanding and elaborating design requirements; and second, how to optimize HMT configurations amid inherent ambiguities in human expert evaluations. To address these gaps, we propose a novel dual-paradigm Hybrid-Augmented Intelligence (HAI) framework that integrates Cognitive Computing (CC-HAI) with Human-in-the-Loop (HITL-HAI) mechanisms. Our key contributions include a CC-HAI–based cognitive teammate mechanism that uses structured prompt engineering to transform LLMs into domain-specialized roles, facilitating the formation of collaborative HMTs; and an HITL-HAI uncertainty mitigation method that employs a Z-number-enhanced cloud modeling approach to manage subjective uncertainties in expert assessments and support robust team configuration. The framework is validated through multi-domain case studies spanning smart home systems, smart cockpits, medical devices, and baby products. Extensive experiments demonstrate its effectiveness in terms of team performance, error reduction, cross-domain generalizability, and decision-making superiority. This research provides a replicable paradigm for deploying LLMs as cognitive collaborators in collaborative design ecosystems, contributing to both theory and methodology in human–machine team intelligence.</div></div>","PeriodicalId":16227,"journal":{"name":"Journal of Manufacturing Systems","volume":"83 ","pages":"Pages 306-321"},"PeriodicalIF":14.2,"publicationDate":"2025-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145118788","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-09-23DOI: 10.1016/j.jmsy.2025.09.008
Yuhan Yuan , Yanfeng Han , Ke Xiao , Zhongying Xu , Xiaomo Jiang
Reducer degradation in robot joints causes excessive vibrations, affecting product quality. Remaining useful life (RUL) prediction of reducers using in-situ signals can avoid robot disassembly and reduces production downtime. However, in-situ signals are more complex than experimental data due to transient robot operations and industrial noise. To address this challenge, an in-situ RUL prediction method via lightweight Multiscale Attention Deep Network (MSADN) and current signal is proposed. First, the full life cycle of harmonic reducer in-situ signals is collected to build a dataset. Subsequently, the MSADN model is employed for RUL prediction. Within MSADN, a multiscale feature extraction (MSFE) module is designed to capture multiscale information from in-situ signals, while a downsampling filter layer (DFL) is incorporated to expand the receptive field. Finally, a novel evaluation metric, Epoch Toleration Accuracy (ETA), alongside other standard evaluation indicators, is introduced to assess RUL prediction performance. Experimental studies on industrial robot datasets and rolling bearing datasets demonstrate the effectiveness and superiority of the proposed MSADN, and two ablation studies validate the necessity of each MSADN component.
{"title":"Remaining useful life prediction for the harmonic reducer of industrial robots via in-situ current signal and lightweight multiscale attention deep networks","authors":"Yuhan Yuan , Yanfeng Han , Ke Xiao , Zhongying Xu , Xiaomo Jiang","doi":"10.1016/j.jmsy.2025.09.008","DOIUrl":"10.1016/j.jmsy.2025.09.008","url":null,"abstract":"<div><div>Reducer degradation in robot joints causes excessive vibrations, affecting product quality. Remaining useful life (RUL) prediction of reducers using in-situ signals can avoid robot disassembly and reduces production downtime. However, in-situ signals are more complex than experimental data due to transient robot operations and industrial noise. To address this challenge, an in-situ RUL prediction method via lightweight Multiscale Attention Deep Network (MSADN) and current signal is proposed. First, the full life cycle of harmonic reducer in-situ signals is collected to build a dataset. Subsequently, the MSADN model is employed for RUL prediction. Within MSADN, a multiscale feature extraction (MSFE) module is designed to capture multiscale information from in-situ signals, while a downsampling filter layer (DFL) is incorporated to expand the receptive field. Finally, a novel evaluation metric, Epoch Toleration Accuracy (ETA), alongside other standard evaluation indicators, is introduced to assess RUL prediction performance. Experimental studies on industrial robot datasets and rolling bearing datasets demonstrate the effectiveness and superiority of the proposed MSADN, and two ablation studies validate the necessity of each MSADN component.</div></div>","PeriodicalId":16227,"journal":{"name":"Journal of Manufacturing Systems","volume":"83 ","pages":"Pages 322-336"},"PeriodicalIF":14.2,"publicationDate":"2025-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145118789","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
To reduce the environmental impacts and resource utilization of End-of-Life (EOL) product recycling, it is imperative to achieve the high efficiency of EOL product recycling and reutilization, including disassembly. However, the disassembly of EOL products is being faced with huge challenges due to the uncertainties of EOL product recycling and dynamic disassembly requirements. Therefore, this paper proposes a digital twin (DT)-assisted multi-agent human-robot collaboration (HRC) disassembly system with multi-scenario data simulations to achieve multi-agent disassembly operations and process optimization. In addition, the dynamic disassembly structure based on dynamic Time Petri Net (TPN) model represents the real-time disassembly information and associated disassembly relationships, which incorporates the digital twin technology to simulate the application environment of HRC disassembly operations. By integrating the multi-agent Dueling-Double deep Q-learning network (MADDQN) algorithm to determine the optimal disassembly sequence and associated task strategy in the DT-assisted HRC disassembly platform. Similarly, it is essential to evaluate the performance of the proposed algorithm for multi-task disassembly planning based on HRC disassembly operations. By conducting an in-depth analysis of the NEV-P50 battery pack from the Weilai ES8 as a case study, the practical implementation of the MADDQN algorithm is demonstrated to optimize the dynamic disassembly sequence and uncertain task allocation with DT data, which provides an effective and flexible approach to the complex disassembly tasks in multi-scenario HRC disassembly processes.
为了减少报废产品回收对环境的影响和资源的利用，必须实现报废产品的高效回收和再利用，包括拆解。然而，由于EOL产品回收的不确定性和拆解需求的动态性，EOL产品的拆解面临着巨大的挑战。为此，本文提出了一种数字孪生（DT）辅助的多智能体人机协作（HRC）拆卸系统，通过多场景数据仿真实现多智能体拆卸操作和工艺优化。此外，基于动态时间Petri网（TPN）模型的动态拆卸结构表示了实时拆卸信息和相关拆卸关系，并结合数字孪生技术模拟了HRC拆卸操作的应用环境。通过集成多智能体duelling - double deep Q-learning network （MADDQN）算法，确定dt辅助HRC拆卸平台的最优拆卸顺序和相关任务策略。同样，对基于HRC拆卸操作的多任务拆卸规划算法的性能进行评估也是必要的。通过对蔚来ES8新能源汽车p50电池组的深入分析，以实际应用为例，展示了基于DT数据的madqn算法对动态拆卸顺序和不确定任务分配的优化，为多场景HRC拆卸过程中复杂的拆卸任务提供了一种有效而灵活的方法。
{"title":"Multi-scenario digital twin-driven human-robot collaboration multi-task disassembly process planning based on dynamic time petri-net and heterogeneous multi-agent double deep Q-learning network","authors":"Jinhua Xiao , Zhiwen Zhang , Sergio Terzi , Fei Tao , Nabil Anwer , Benoit Eynard","doi":"10.1016/j.jmsy.2025.09.011","DOIUrl":"10.1016/j.jmsy.2025.09.011","url":null,"abstract":"<div><div>To reduce the environmental impacts and resource utilization of End-of-Life (EOL) product recycling, it is imperative to achieve the high efficiency of EOL product recycling and reutilization, including disassembly. However, the disassembly of EOL products is being faced with huge challenges due to the uncertainties of EOL product recycling and dynamic disassembly requirements. Therefore, this paper proposes a digital twin (DT)-assisted multi-agent human-robot collaboration (HRC) disassembly system with multi-scenario data simulations to achieve multi-agent disassembly operations and process optimization. In addition, the dynamic disassembly structure based on dynamic Time Petri Net (TPN) model represents the real-time disassembly information and associated disassembly relationships, which incorporates the digital twin technology to simulate the application environment of HRC disassembly operations. By integrating the multi-agent Dueling-Double deep Q-learning network (MADDQN) algorithm to determine the optimal disassembly sequence and associated task strategy in the DT-assisted HRC disassembly platform. Similarly, it is essential to evaluate the performance of the proposed algorithm for multi-task disassembly planning based on HRC disassembly operations. By conducting an in-depth analysis of the NEV-P50 battery pack from the Weilai ES8 as a case study, the practical implementation of the MADDQN algorithm is demonstrated to optimize the dynamic disassembly sequence and uncertain task allocation with DT data, which provides an effective and flexible approach to the complex disassembly tasks in multi-scenario HRC disassembly processes.</div></div>","PeriodicalId":16227,"journal":{"name":"Journal of Manufacturing Systems","volume":"83 ","pages":"Pages 284-305"},"PeriodicalIF":14.2,"publicationDate":"2025-09-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145096557","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-09-17DOI: 10.1016/j.jmsy.2025.09.006
Yu Zhang , Zeqiang Zhang , Feng Chu , Yanqing Zeng , Lei Guo , Zongxing He
Robotic disassembly lines play a pivotal role in remanufacturing by enabling automated operations. In two-sided disassembly scenarios involving large-scale products such as automobiles, their high load capacity significantly reduces the labor intensity of manual disassembly and eliminates the need for lifting equipment, thereby streamlining the process flow. When equipped with mobility systems, 7-axis robots can flexibly switch between multiple workstations, facilitating both rapid adaptation to process changes and precise execution of spatially heterogeneous disassembly tasks. However, despite these advantages, systematic research on the integration of mobile disassembly robots within disassembly line applications remains limited. To address this gap, this study integrates 7-axis mobile robots into two-sided disassembly lines and models the system using both mixed-integer programming and constraint programming approaches. The proposed models aim to minimize construction costs and ensure balanced workload distribution across stations. A novel constraint programming-based lexicographic-Pareto approach is developed to solve the resulting multi-objective optimization problem, this method is capable of generating verified Pareto frontiers for small-scale instances and providing high-quality approximate Pareto solution sets for large-scale problems. In the numerical experiments, a sensitivity analysis of key algorithm parameters is first conducted to achieve a balance between computational efficiency and solution quality. Subsequently, the proposed method is benchmarked against nine existing algorithms across twenty datasets to validate its effectiveness. Its practical feasibility is further demonstrated through an application to the disassembly of drum washing machines. The results show that, compared to conventional fixed-robot disassembly lines without cross-station coordination, the mobile robot configuration achieves a 10.7% reduction in total cost and a 66.7% improvement in robot workload balance, offering a promising pathway for advancing remanufacturing practices.
{"title":"A constraint programming-based lexicographic-Pareto approach for balancing two-sided robotic disassembly lines with 7-axis robots","authors":"Yu Zhang , Zeqiang Zhang , Feng Chu , Yanqing Zeng , Lei Guo , Zongxing He","doi":"10.1016/j.jmsy.2025.09.006","DOIUrl":"10.1016/j.jmsy.2025.09.006","url":null,"abstract":"<div><div>Robotic disassembly lines play a pivotal role in remanufacturing by enabling automated operations. In two-sided disassembly scenarios involving large-scale products such as automobiles, their high load capacity significantly reduces the labor intensity of manual disassembly and eliminates the need for lifting equipment, thereby streamlining the process flow. When equipped with mobility systems, 7-axis robots can flexibly switch between multiple workstations, facilitating both rapid adaptation to process changes and precise execution of spatially heterogeneous disassembly tasks. However, despite these advantages, systematic research on the integration of mobile disassembly robots within disassembly line applications remains limited. To address this gap, this study integrates 7-axis mobile robots into two-sided disassembly lines and models the system using both mixed-integer programming and constraint programming approaches. The proposed models aim to minimize construction costs and ensure balanced workload distribution across stations. A novel constraint programming-based lexicographic-Pareto approach is developed to solve the resulting multi-objective optimization problem, this method is capable of generating verified Pareto frontiers for small-scale instances and providing high-quality approximate Pareto solution sets for large-scale problems. In the numerical experiments, a sensitivity analysis of key algorithm parameters is first conducted to achieve a balance between computational efficiency and solution quality. Subsequently, the proposed method is benchmarked against nine existing algorithms across twenty datasets to validate its effectiveness. Its practical feasibility is further demonstrated through an application to the disassembly of drum washing machines. The results show that, compared to conventional fixed-robot disassembly lines without cross-station coordination, the mobile robot configuration achieves a 10.7% reduction in total cost and a 66.7% improvement in robot workload balance, offering a promising pathway for advancing remanufacturing practices.</div></div>","PeriodicalId":16227,"journal":{"name":"Journal of Manufacturing Systems","volume":"83 ","pages":"Pages 235-251"},"PeriodicalIF":14.2,"publicationDate":"2025-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145096558","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-09-17DOI: 10.1016/j.jmsy.2025.09.009
Teng Zhang , Fangyu Peng , Zhao Yang , Xiaowei Tang , Jiangmiao Yuan , Rong Yan
Robotic machining has become another important machining paradigm after CNC machine tools. However, robot error has always been an important constraint in its progress towards high quality demand scenarios due to characteristics such as weak rigidity and pose dependence. Numerous scholars have carried out rich work around errors in robotic machining systems, and these studies have achieved excellent results in robot localization, trajectory continuous motion, and machining operations. However, due to the complexity of the robot machining system, the robot error has differentiated performance at different stages, and it is difficult to guarantee the global accuracy of the robot by focusing on and controlling a certain kind of error in a discrete manner. For this reason, a digital twin-driven staged error prediction and compensation framework for the whole robot machining process is constructed. In this framework, the whole process of robot machining is divided into three stages with significant differences: point planning, trajectory planning and material removal. And the error prediction function block in each stage is constructed for the error characteristics (distribution skew, error step, spatial-temporal coupling). For error compensation, a staged error compensation strategy is constructed from three aspects: offline point position, robot body and external three-axis platform, respectively. The constructed system was case-validated in the robotic machining of curved parts. All stages of the error prediction models show high prediction accuracy, and the excellent performance of the staged prediction models is verified by comparing with the classical prediction models. For the error compensation, the designed system is utilized to ensure that the robotic machining system provides a double guarantee on the robot end and the machining quality, the point position absolute error is controlled at 0.109 mm, the orientation error is controlled at 0.028°, the trajectory position error is controlled at 0.067 mm, the orientation error is controlled at 0.031°, and the final part machining error is controlled at 0.036 mm, which is almost approximates the repeatable positioning accuracy of the robot. The proposed framework realizes the system-level sensing and control of the robot machining system error, and provides a unified system framework for the subsequent research of related unit methods, which is conducive to promoting the development of robot machining to high-quality requirement scenarios.
{"title":"Digital twin-driven staged error prediction and compensation framework for the whole process of robotic machining","authors":"Teng Zhang , Fangyu Peng , Zhao Yang , Xiaowei Tang , Jiangmiao Yuan , Rong Yan","doi":"10.1016/j.jmsy.2025.09.009","DOIUrl":"10.1016/j.jmsy.2025.09.009","url":null,"abstract":"<div><div>Robotic machining has become another important machining paradigm after CNC machine tools. However, robot error has always been an important constraint in its progress towards high quality demand scenarios due to characteristics such as weak rigidity and pose dependence. Numerous scholars have carried out rich work around errors in robotic machining systems, and these studies have achieved excellent results in robot localization, trajectory continuous motion, and machining operations. However, due to the complexity of the robot machining system, the robot error has differentiated performance at different stages, and it is difficult to guarantee the global accuracy of the robot by focusing on and controlling a certain kind of error in a discrete manner. For this reason, a digital twin-driven staged error prediction and compensation framework for the whole robot machining process is constructed. In this framework, the whole process of robot machining is divided into three stages with significant differences: point planning, trajectory planning and material removal. And the error prediction function block in each stage is constructed for the error characteristics (distribution skew, error step, spatial-temporal coupling). For error compensation, a staged error compensation strategy is constructed from three aspects: offline point position, robot body and external three-axis platform, respectively. The constructed system was case-validated in the robotic machining of curved parts. All stages of the error prediction models show high prediction accuracy, and the excellent performance of the staged prediction models is verified by comparing with the classical prediction models. For the error compensation, the designed system is utilized to ensure that the robotic machining system provides a double guarantee on the robot end and the machining quality, the point position absolute error is controlled at 0.109 mm, the orientation error is controlled at 0.028°, the trajectory position error is controlled at 0.067 mm, the orientation error is controlled at 0.031°, and the final part machining error is controlled at 0.036 mm, which is almost approximates the repeatable positioning accuracy of the robot. The proposed framework realizes the system-level sensing and control of the robot machining system error, and provides a unified system framework for the subsequent research of related unit methods, which is conducive to promoting the development of robot machining to high-quality requirement scenarios.</div></div>","PeriodicalId":16227,"journal":{"name":"Journal of Manufacturing Systems","volume":"83 ","pages":"Pages 252-283"},"PeriodicalIF":14.2,"publicationDate":"2025-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145096556","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-09-16DOI: 10.1016/j.jmsy.2025.08.016
Chen Li, Qing Chang
Flexible Smart Manufacturing Systems (FSMS) are critical to achieving mass customization and operational agility under Industry 4.0. However, planning effective FSMS configurations remains challenging due to fluctuating market demands, heterogeneous system components, complex interdependencies, and the need to optimize resource utilization. Conventional planning methods often require predefined line configurations and lack adaptability, scalability, and awareness of dynamic system properties. This paper presents a novel Hybrid Graph-Diffusion Based Planning Framework that integrates generative AI with system-theoretic modeling to autonomously generate optimal FSMS configurations based on different market demands. Specifically, we introduce a system model-embedded Heterogeneous Graph (HG) to represent the structure and properties of an FSMS and infuse it within a system property-tailored diffusion model to generate reconfigurable plan configurations. The final system property-guided refinement guarantees that the final plan configuration is optimal in both demand satisfaction and resource use. Furthermore, our ablation studies validate that our framework significantly outperforms conventional approaches in both demand satisfaction and resource efficiency. Furthermore, our ablation studies validate the effectiveness of the system property guidance and HG-based representation in enhancing planning feasibility, robustness, and adaptability.
{"title":"Generative AI-powered planning: A hybrid graph-diffusion approach for demand-driven flexible manufacturing systems","authors":"Chen Li, Qing Chang","doi":"10.1016/j.jmsy.2025.08.016","DOIUrl":"10.1016/j.jmsy.2025.08.016","url":null,"abstract":"<div><div>Flexible Smart Manufacturing Systems (FSMS) are critical to achieving mass customization and operational agility under Industry 4.0. However, planning effective FSMS configurations remains challenging due to fluctuating market demands, heterogeneous system components, complex interdependencies, and the need to optimize resource utilization. Conventional planning methods often require predefined line configurations and lack adaptability, scalability, and awareness of dynamic system properties. This paper presents a novel Hybrid Graph-Diffusion Based Planning Framework that integrates generative AI with system-theoretic modeling to autonomously generate optimal FSMS configurations based on different market demands. Specifically, we introduce a system model-embedded Heterogeneous Graph (HG) to represent the structure and properties of an FSMS and infuse it within a system property-tailored diffusion model to generate reconfigurable plan configurations. The final system property-guided refinement guarantees that the final plan configuration is optimal in both demand satisfaction and resource use. Furthermore, our ablation studies validate that our framework significantly outperforms conventional approaches in both demand satisfaction and resource efficiency. Furthermore, our ablation studies validate the effectiveness of the system property guidance and HG-based representation in enhancing planning feasibility, robustness, and adaptability.</div></div>","PeriodicalId":16227,"journal":{"name":"Journal of Manufacturing Systems","volume":"83 ","pages":"Pages 175-195"},"PeriodicalIF":14.2,"publicationDate":"2025-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145096553","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号