Robotics and Computer-integrated Manufacturing最新文献_第8页

From perception to precision: Vision-based mobile robotic manipulation for assembly screwdriving 从感知到精确：基于视觉的移动机器人操作装配螺丝

IF 11.4 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Robotics and Computer-integrated Manufacturing

Pub Date : 2025-09-30 DOI: 10.1016/j.rcim.2025.103148

Aleksandar Stefanov, Miha Zorman, Sebastjan Šlajpah, Janez Podobnik, Matjaž Mihelj, Marko Munih

Flexible manufacturing demands automation that is both precise and adaptable. However, tasks such as screwdriving are typically automated using costly, rigid robotic cells, making this approach impractical for low-volume, high-mix production. As a scalable solution, mobile manipulators offer a flexible alternative, but achieving the required precision for screwdriving remains challenging due to localization uncertainties. This paper addresses these limitations by presenting a vision-guided mobile robotic manipulation system that performs high-precision screwdriving using only monocular RGB imagery. The proposed pipeline integrates stationary and onboard cameras with perception algorithms for object identification and segmentation, pose estimation, and CAD-based screw hole localization, compensating for base misalignment and object placement variability. Experimental validation using ISO 9283 standard’s metrics demonstrates a translational accuracy between 0.21 mm and 0.50 mm across multiple screw positions. Additionally, the system achieves angular estimation errors as low as 0.07°to 0.20°, verifying its capability for sub-degree precision in orientation estimation. In 50 independent experiments involving a total of 400 screw insertions, the system achieved a 100 % success rate, confirming its reliability in practical conditions. These results confirm the feasibility of using RGB-only vision for precision screwdriving and highlight the mobile manipulation system’s scalability for real-world semi-structured manufacturing environments.

柔性制造要求自动化既精确又适应性强。然而，像螺丝刀这样的任务通常是自动化的，使用昂贵的、刚性的机器人单元，这使得这种方法不适合小批量、高混合的生产。作为一种可扩展的解决方案，移动机械手提供了一种灵活的选择，但由于定位的不确定性，实现所需的螺丝刀精度仍然具有挑战性。本文通过提出一种视觉引导的移动机器人操作系统来解决这些限制，该系统仅使用单目RGB图像执行高精度螺丝刀。该管道集成了固定式和机载摄像机，以及用于物体识别和分割、姿态估计和基于cad的螺钉孔定位的感知算法，补偿了基座错位和物体放置的可变性。使用ISO 9283标准度量的实验验证表明，在多个螺钉位置上的平移精度在0.21 mm至0.50 mm之间。此外，该系统的角度估计误差低至0.07°至0.20°，验证了其在定向估计中的亚度精度能力。在50次独立实验中，共进行了400次螺钉置入，成功率达到100%，证实了该系统在实际条件下的可靠性。这些结果证实了仅使用rgb视觉进行精密螺丝刀的可行性，并突出了移动操作系统在现实世界半结构化制造环境中的可扩展性。

{"title":"From perception to precision: Vision-based mobile robotic manipulation for assembly screwdriving","authors":"Aleksandar Stefanov, Miha Zorman, Sebastjan Šlajpah, Janez Podobnik, Matjaž Mihelj, Marko Munih","doi":"10.1016/j.rcim.2025.103148","DOIUrl":"10.1016/j.rcim.2025.103148","url":null,"abstract":"<div><div>Flexible manufacturing demands automation that is both precise and adaptable. However, tasks such as screwdriving are typically automated using costly, rigid robotic cells, making this approach impractical for low-volume, high-mix production. As a scalable solution, mobile manipulators offer a flexible alternative, but achieving the required precision for screwdriving remains challenging due to localization uncertainties. This paper addresses these limitations by presenting a vision-guided mobile robotic manipulation system that performs high-precision screwdriving using only monocular RGB imagery. The proposed pipeline integrates stationary and onboard cameras with perception algorithms for object identification and segmentation, pose estimation, and CAD-based screw hole localization, compensating for base misalignment and object placement variability. Experimental validation using ISO 9283 standard’s metrics demonstrates a translational accuracy between 0.21 mm and 0.50 mm across multiple screw positions. Additionally, the system achieves angular estimation errors as low as 0.07°to 0.20°, verifying its capability for sub-degree precision in orientation estimation. In 50 independent experiments involving a total of 400 screw insertions, the system achieved a 100 % success rate, confirming its reliability in practical conditions. These results confirm the feasibility of using RGB-only vision for precision screwdriving and highlight the mobile manipulation system’s scalability for real-world semi-structured manufacturing environments.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"98 ","pages":"Article 103148"},"PeriodicalIF":11.4,"publicationDate":"2025-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145220393","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Intelligent support-free additive manufacturing path planning via method library and neural network 基于方法库和神经网络的智能无支撑增材制造路径规划

IF 11.4 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Robotics and Computer-integrated Manufacturing

Pub Date : 2025-09-30 DOI: 10.1016/j.rcim.2025.103156

Zhengren Tong , Lai Xu , Xianglong Li , Chen Yang , Qinfeng Wang , Hongyao Shen

Support-free additive manufacturing achieves self-supporting fabrication by adjusting the platform’s posture, effectively reducing material waste and simplifying the post-processing stage. However, the diversity of industrial part geometries requires different approaches to plan the robotic manufacturing paths. Traditional approaches to selecting support-free manufacturing path planning methods rely heavily on expert knowledge. This paper proposes an intelligent path planning system based on a method library. The system utilizes a method library composed of seven approaches to achieve support-free additive manufacturing path planning of various types of parts. The additive manufacturing strategy matching neural network (AMMatcher) is employed to match the optimal path planning method from the method library to a given model and to identify its base surface. AMMatcher can analyze the multi-scale features of the model and leverage a cross-task attention mechanism to propagate classification features into the segmentation task, thereby improving network performance. A newly proposed support-free additive manufacturing model dataset (SFAMDataset) is used to evaluate the performance of AMMatcher and typical samples are validated through fabrication experiments on three different manufacturing platforms. Experimental results demonstrate that AMMatcher effectively identifies suitable manufacturing strategies for various model types and exhibits strong adaptability across different manufacturing platforms.

无支撑增材制造通过调整平台姿态实现自支撑制造，有效减少材料浪费，简化后处理阶段。然而，工业零件几何形状的多样性需要不同的方法来规划机器人制造路径。传统的选择无支持制造路径规划方法严重依赖于专家知识。提出了一种基于方法库的智能路径规划系统。该系统利用由7种方法组成的方法库实现了各类零件的无支撑增材制造路径规划。利用增材制造策略匹配神经网络（AMMatcher）将方法库中的最优路径规划方法与给定模型进行匹配，并识别其基面。AMMatcher可以分析模型的多尺度特征，并利用跨任务注意机制将分类特征传播到分割任务中，从而提高网络性能。利用新提出的无支撑增材制造模型数据集（SFAMDataset）来评估AMMatcher的性能，并通过三种不同制造平台的制造实验对典型样本进行了验证。实验结果表明，AMMatcher能够有效地识别出适合不同模型类型的制造策略，并在不同制造平台上表现出较强的适应性。

{"title":"Intelligent support-free additive manufacturing path planning via method library and neural network","authors":"Zhengren Tong , Lai Xu , Xianglong Li , Chen Yang , Qinfeng Wang , Hongyao Shen","doi":"10.1016/j.rcim.2025.103156","DOIUrl":"10.1016/j.rcim.2025.103156","url":null,"abstract":"<div><div>Support-free additive manufacturing achieves self-supporting fabrication by adjusting the platform’s posture, effectively reducing material waste and simplifying the post-processing stage. However, the diversity of industrial part geometries requires different approaches to plan the robotic manufacturing paths. Traditional approaches to selecting support-free manufacturing path planning methods rely heavily on expert knowledge. This paper proposes an intelligent path planning system based on a method library. The system utilizes a method library composed of seven approaches to achieve support-free additive manufacturing path planning of various types of parts. The additive manufacturing strategy matching neural network (AMMatcher) is employed to match the optimal path planning method from the method library to a given model and to identify its base surface. AMMatcher can analyze the multi-scale features of the model and leverage a cross-task attention mechanism to propagate classification features into the segmentation task, thereby improving network performance. A newly proposed support-free additive manufacturing model dataset (SFAMDataset) is used to evaluate the performance of AMMatcher and typical samples are validated through fabrication experiments on three different manufacturing platforms. Experimental results demonstrate that AMMatcher effectively identifies suitable manufacturing strategies for various model types and exhibits strong adaptability across different manufacturing platforms.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"98 ","pages":"Article 103156"},"PeriodicalIF":11.4,"publicationDate":"2025-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145220967","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Simultaneous high transparency and robust stability-oriented Physical Human-Robot Interaction using an Interaction Intention Filter and a vibration observer 基于交互意图滤波器和振动观测器的高透明鲁棒稳定的人机交互

IF 11.4 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Robotics and Computer-integrated Manufacturing

Pub Date : 2025-09-30 DOI: 10.1016/j.rcim.2025.103150

Junsheng Huang, Mingxing Yuan, Xuebo Zhang

Physical Human-Robot Interaction (pHRI) systems equipped with compliant admittance controllers typically utilize F/T sensors to capture the forces applied by the operator. However, the impedance force feedback generated by the robot’s motion and the impedance of the human hand can significantly distort the intentional forces. This distortion can lead to vibrations that compromise both interaction transparency and stability. To address this issue, we propose a variable admittance control strategy that incorporates an Interaction Intention Filter (IIF) and an Enhanced Time-Domain Vibration Observer (ETDVO). We first introduce the concept of the IIF, which is designed based on a frequency-domain analysis of force signals collected from real-world human–robot cooperation tasks. This filter effectively prevents unintended impedance force feedback from being transmitted to the admittance controller. Moreover, to ensure interaction stability across diverse environments, we propose a variable-width time window-based ETDVO for accurately computing the vibration index. By leveraging this index, we introduce a variable admittance control strategy based on exponential mapping, which enables rapid adjustment of the admittance parameters, effectively suppresses vibrations and enhances stability. Finally, the proposed strategy is validated through human–robot cooperative laser tracking experiments conducted on a 7-DoF manipulator. Statistical results from the experiments demonstrate that our approach not only improves interaction transparency but also significantly enhances overall stability. Compared to the stable high-gain admittance controller, the Task Time, Required Energy, and Mean Force are reduced by over 10%, 54%, 58%, respectively.

配备符合导纳控制器的物理人机交互（pHRI）系统通常使用F/T传感器来捕获操作员施加的力。然而，机器人运动产生的阻抗力反馈和人手的阻抗会显著地扭曲意向性力。这种扭曲会导致振动，损害相互作用的透明度和稳定性。为了解决这个问题，我们提出了一种可变导纳控制策略，该策略结合了交互意图滤波器（IIF）和增强时域振动观测器（ETDVO）。我们首先介绍了IIF的概念，它是基于从现实世界的人机合作任务中收集的力信号的频域分析而设计的。该滤波器有效地防止意外的阻抗力反馈被传输到导纳控制器。此外，为了确保在不同环境下的相互作用稳定性，我们提出了一种基于变宽时窗的ETDVO来精确计算振动指数。利用这一指标，我们引入了一种基于指数映射的可变导纳控制策略，该策略可以快速调整导纳参数，有效地抑制振动并提高稳定性。最后，在一个7自由度机械臂上进行了人机协同激光跟踪实验，验证了所提策略的有效性。实验统计结果表明，我们的方法不仅提高了交互透明度，而且显著提高了整体稳定性。与稳定的高增益导纳控制器相比，任务时间、所需能量和平均力分别降低了10%、54%和58%以上。

{"title":"Simultaneous high transparency and robust stability-oriented Physical Human-Robot Interaction using an Interaction Intention Filter and a vibration observer","authors":"Junsheng Huang, Mingxing Yuan, Xuebo Zhang","doi":"10.1016/j.rcim.2025.103150","DOIUrl":"10.1016/j.rcim.2025.103150","url":null,"abstract":"<div><div>Physical Human-Robot Interaction (pHRI) systems equipped with compliant admittance controllers typically utilize F/T sensors to capture the forces applied by the operator. However, the impedance force feedback generated by the robot’s motion and the impedance of the human hand can significantly distort the intentional forces. This distortion can lead to vibrations that compromise both interaction transparency and stability. To address this issue, we propose a variable admittance control strategy that incorporates an Interaction Intention Filter (IIF) and an Enhanced Time-Domain Vibration Observer (ETDVO). We first introduce the concept of the IIF, which is designed based on a frequency-domain analysis of force signals collected from real-world human–robot cooperation tasks. This filter effectively prevents unintended impedance force feedback from being transmitted to the admittance controller. Moreover, to ensure interaction stability across diverse environments, we propose a variable-width time window-based ETDVO for accurately computing the vibration index. By leveraging this index, we introduce a variable admittance control strategy based on exponential mapping, which enables rapid adjustment of the admittance parameters, effectively suppresses vibrations and enhances stability. Finally, the proposed strategy is validated through human–robot cooperative laser tracking experiments conducted on a 7-DoF manipulator. Statistical results from the experiments demonstrate that our approach not only improves interaction transparency but also significantly enhances overall stability. Compared to the stable high-gain admittance controller, the <em>Task Time</em>, <em>Required Energy</em>, and <em>Mean Force</em> are reduced by over 10%, 54%, 58%, respectively.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"98 ","pages":"Article 103150"},"PeriodicalIF":11.4,"publicationDate":"2025-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145220388","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Modeling and compensation of measurement errors in hand-eye system for heavy-load industrial robots with line laser sensor 线激光传感器重载工业机器人手眼系统测量误差建模与补偿

IF 11.4 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Robotics and Computer-integrated Manufacturing

Pub Date : 2025-09-30 DOI: 10.1016/j.rcim.2025.103155

Xiaoyu Guo , Bao Zhu , Meng Chi , Chen Liu , Yanding Wei , Qiang Fang

During the continuous scanning process in which a heavy-load robot carries a line laser sensor, measurement accuracy is susceptible to the influence of both geometric errors and joint deformations. Traditional elastogeometric error compensation methods often rely heavily on the calibration accuracy of external measurement systems, which limits their flexibility and precision in on-site applications. To address this limitation, this study proposed Multi-Set Cohesive Calibration (MSCC), a method that eliminates the need for high-precision external system calibration before parameter identification. The MSCC integrated robot geometric errors, compliance errors, and extrinsic parameter errors into a unified error model, solving them collaboratively using multi-configuration measurement data, thereby enhancing the stability and adaptability of the calibration system. Furthermore, to address the high-dimensional and strongly coupled parameter identification problem, a three-stage hybrid optimization algorithm called the Exploration-Annealing-LM (EALM) algorithm was introduced to improve the convergence and global search capability during parameter estimation. The results demonstrated that, in online measurement applications for large structural components, the proposed method achieves an average measurement error of 0.0545 mm and a maximum error of 0.1296 mm, representing reductions of 84.36% and 78.31%, respectively, compared to the uncompensated case.

在重载机器人携带直线激光传感器的连续扫描过程中，测量精度容易受到几何误差和关节变形的影响。传统的弹性几何误差补偿方法往往严重依赖于外部测量系统的校准精度，这限制了其在现场应用中的灵活性和精度。为了解决这一限制，本研究提出了多集内聚校准（MSCC）方法，该方法在参数识别之前不需要高精度的外部系统校准。MSCC将机器人几何误差、柔度误差和外部参数误差整合到一个统一的误差模型中，利用多组态测量数据协同求解，从而增强了标定系统的稳定性和适应性。此外，为了解决高维强耦合参数辨识问题，引入了一种三阶段混合优化算法——探索-退火- lm （EALM）算法，提高了参数估计过程中的收敛性和全局搜索能力。结果表明，在大型结构件在线测量应用中，该方法的平均测量误差为0.0545 mm，最大误差为0.1296 mm，与未补偿情况相比，分别减小了84.36%和78.31%。

{"title":"Modeling and compensation of measurement errors in hand-eye system for heavy-load industrial robots with line laser sensor","authors":"Xiaoyu Guo , Bao Zhu , Meng Chi , Chen Liu , Yanding Wei , Qiang Fang","doi":"10.1016/j.rcim.2025.103155","DOIUrl":"10.1016/j.rcim.2025.103155","url":null,"abstract":"<div><div>During the continuous scanning process in which a heavy-load robot carries a line laser sensor, measurement accuracy is susceptible to the influence of both geometric errors and joint deformations. Traditional elastogeometric error compensation methods often rely heavily on the calibration accuracy of external measurement systems, which limits their flexibility and precision in on-site applications. To address this limitation, this study proposed Multi-Set Cohesive Calibration (MSCC), a method that eliminates the need for high-precision external system calibration before parameter identification. The MSCC integrated robot geometric errors, compliance errors, and extrinsic parameter errors into a unified error model, solving them collaboratively using multi-configuration measurement data, thereby enhancing the stability and adaptability of the calibration system. Furthermore, to address the high-dimensional and strongly coupled parameter identification problem, a three-stage hybrid optimization algorithm called the Exploration-Annealing-LM (EALM) algorithm was introduced to improve the convergence and global search capability during parameter estimation. The results demonstrated that, in online measurement applications for large structural components, the proposed method achieves an average measurement error of 0.0545 mm and a maximum error of 0.1296 mm, representing reductions of 84.36% and 78.31%, respectively, compared to the uncompensated case.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"98 ","pages":"Article 103155"},"PeriodicalIF":11.4,"publicationDate":"2025-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145220389","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A mixed reality-assisted scene-centric robot programming approach for human–robot collaborative manufacturing 面向人机协同制造的混合现实辅助场景中心机器人编程方法

IF 11.4 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Robotics and Computer-integrated Manufacturing

Pub Date : 2025-09-29 DOI: 10.1016/j.rcim.2025.103146

Yue Yin , Junming Fan , Ang Liu , Pai Zheng

While mass personalization manufacturing paradigm increasingly requires robots to handle complex and variable tasks, traditional robot-centric programming methods remain constrained by their expert-dependent nature and lack of adaptability. To address these limitations, this research proposes a scene-centric robot programming approach using MR-assisted interactive 3D segmentation, where operators naturally manipulate the digital twin (DT) of real-world objects to control the robot, rather than considering cumbersome end-effector programming. This framework combines Segment Anything Model (SAM) and 3D Gaussian Splatting (3DGS) for cost-effective, zero-shot, and flexible scene reconstruction and segmentation. Scale consistency and multi-coordinate calibration ensure seamless MR-driven interaction and robot execution. Finally, experimental results verify improved segmentation accuracy and computational efficiency, particularly in cluttered industrial environments, while case studies validate the method’s feasibility for real-world implementation. This research illustrates a promising human–robot collaborative manufacturing paradigm where virtual scene editing directly informs robot actions, demonstrating a novel MR-assisted interaction method beyond low-level robot movement control.

随着大规模个性化制造模式越来越多地要求机器人处理复杂多变的任务，传统的以机器人为中心的编程方法仍然受到其依赖专家的性质和缺乏适应性的限制。为了解决这些限制，本研究提出了一种以场景为中心的机器人编程方法，使用磁共振辅助的交互式3D分割，其中操作员自然地操纵现实世界对象的数字孪生（DT）来控制机器人，而不是考虑繁琐的末端执行器编程。该框架结合了分段任意模型（SAM）和3D高斯喷溅（3DGS），具有成本效益，零拍摄和灵活的场景重建和分割。尺度一致性和多坐标校准确保了核磁共振驱动的无缝交互和机器人执行。最后，实验结果验证了分割精度和计算效率的提高，特别是在混乱的工业环境中，而案例研究验证了该方法在现实世界实现的可行性。本研究展示了一个有前途的人机协作制造范例，其中虚拟场景编辑直接通知机器人动作，展示了一种超越低级机器人运动控制的新型磁共振辅助交互方法。

{"title":"A mixed reality-assisted scene-centric robot programming approach for human–robot collaborative manufacturing","authors":"Yue Yin , Junming Fan , Ang Liu , Pai Zheng","doi":"10.1016/j.rcim.2025.103146","DOIUrl":"10.1016/j.rcim.2025.103146","url":null,"abstract":"<div><div>While mass personalization manufacturing paradigm increasingly requires robots to handle complex and variable tasks, traditional robot-centric programming methods remain constrained by their expert-dependent nature and lack of adaptability. To address these limitations, this research proposes a scene-centric robot programming approach using MR-assisted interactive 3D segmentation, where operators naturally manipulate the digital twin (DT) of real-world objects to control the robot, rather than considering cumbersome end-effector programming. This framework combines Segment Anything Model (SAM) and 3D Gaussian Splatting (3DGS) for cost-effective, zero-shot, and flexible scene reconstruction and segmentation. Scale consistency and multi-coordinate calibration ensure seamless MR-driven interaction and robot execution. Finally, experimental results verify improved segmentation accuracy and computational efficiency, particularly in cluttered industrial environments, while case studies validate the method’s feasibility for real-world implementation. This research illustrates a promising human–robot collaborative manufacturing paradigm where virtual scene editing directly informs robot actions, demonstrating a novel MR-assisted interaction method beyond low-level robot movement control.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"98 ","pages":"Article 103146"},"PeriodicalIF":11.4,"publicationDate":"2025-09-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145220968","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

CCM-FCC: LLM-powered cognition-centered AI agent framework for proactive human-robot collaboration CCM-FCC：基于llm的主动人机协作认知中心AI代理框架

IF 11.4 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Robotics and Computer-integrated Manufacturing

Pub Date : 2025-09-27 DOI: 10.1016/j.rcim.2025.103145

Pengfei Ding , Jie Zhang , Peng Zhang , Hongsen Li , Dexian Wang

Proactive human-robot collaboration (PHRC) primarily relies on predefined rule-based integration of perception, analysis and decision-making into a unified framework, which limits its autonomy and interactivity in dynamic scenarios such as disassembly and assembly. Although AI agent equipped with memory and interaction functions exhibits enhanced adaptability, their task-specific designs result in a lack of holistic cognition, thereby limiting their generalization capability. This paper proposes a Large Language Model (LLM)-powered cognition-centered AI agent framework, which addresses these challenges through the “Cognitive Core Management–Functional Cluster Collaboration” (CCM-FCC) paradigm. Specifically, to enhance the generalization capability of the AI agent, we developed a semantic Chain-of-Thought (CoT) prompt learning-driven cognitive core for predicting key task factors. The semantic CoT prompt learning, which couples task semantics with reasoning logic, empowers the pre-trained LLM to improve the key factors prediction. Subsequently, to ensure centralized management of the cognitive core, we designed a dual-dimensional feature-constrained functional activation module. It extracts task semantic cues from the key factors and autonomously activates functional modules within the AI agent, constrained by task complexity and operator state. Furthermore, a task-semantic-driven functional cluster collaboration module is proposed to generate the optimal collaboration strategy. Finally, a deep reinforcement learning model is constructed to enable the robot to proactively collaborate with the operator for PHRC. The experiments on HRC tasks demonstrates the effectiveness of the proposed method.

主动人机协作（PHRC）主要依赖于预定义的基于规则的感知、分析和决策集成到一个统一的框架中，这限制了其在拆卸和组装等动态场景中的自主性和交互性。虽然具有记忆和交互功能的人工智能智能体具有较强的适应性，但其特定任务的设计导致其缺乏整体认知，从而限制了其泛化能力。本文提出了一个基于大语言模型（LLM）的以认知为中心的人工智能代理框架，该框架通过“认知核心管理-功能集群协作”（CCM-FCC）范式解决了这些挑战。具体而言，为了提高人工智能智能体的泛化能力，我们开发了一个语义思维链（CoT）提示学习驱动的认知核心，用于预测关键任务因素。语义CoT提示学习将任务语义与推理逻辑相结合，使预训练的LLM能够提高关键因素的预测能力。随后，为了保证认知核心的集中管理，我们设计了一个二维特征约束的功能激活模块。它从关键因素中提取任务语义线索，并在任务复杂性和操作员状态的约束下自主激活人工智能代理内的功能模块。在此基础上，提出了一个任务语义驱动的功能集群协同模块来生成最优协同策略。最后，构建了一个深度强化学习模型，使机器人能够主动与操作员协作进行PHRC。在HRC任务上的实验验证了该方法的有效性。

{"title":"CCM-FCC: LLM-powered cognition-centered AI agent framework for proactive human-robot collaboration","authors":"Pengfei Ding , Jie Zhang , Peng Zhang , Hongsen Li , Dexian Wang","doi":"10.1016/j.rcim.2025.103145","DOIUrl":"10.1016/j.rcim.2025.103145","url":null,"abstract":"<div><div>Proactive human-robot collaboration (PHRC) primarily relies on predefined rule-based integration of perception, analysis and decision-making into a unified framework, which limits its autonomy and interactivity in dynamic scenarios such as disassembly and assembly. Although AI agent equipped with memory and interaction functions exhibits enhanced adaptability, their task-specific designs result in a lack of holistic cognition, thereby limiting their generalization capability. This paper proposes a Large Language Model (LLM)-powered cognition-centered AI agent framework, which addresses these challenges through the “Cognitive Core Management–Functional Cluster Collaboration” (CCM-FCC) paradigm. Specifically, to enhance the generalization capability of the AI agent, we developed a semantic Chain-of-Thought (CoT) prompt learning-driven cognitive core for predicting key task factors. The semantic CoT prompt learning, which couples task semantics with reasoning logic, empowers the pre-trained LLM to improve the key factors prediction. Subsequently, to ensure centralized management of the cognitive core, we designed a dual-dimensional feature-constrained functional activation module. It extracts task semantic cues from the key factors and autonomously activates functional modules within the AI agent, constrained by task complexity and operator state. Furthermore, a task-semantic-driven functional cluster collaboration module is proposed to generate the optimal collaboration strategy. Finally, a deep reinforcement learning model is constructed to enable the robot to proactively collaborate with the operator for PHRC. The experiments on HRC tasks demonstrates the effectiveness of the proposed method.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"98 ","pages":"Article 103145"},"PeriodicalIF":11.4,"publicationDate":"2025-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145158950","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Human-aware scheduling for sustainable manufacturing: A review of dynamic job shop scheduling in the era of Industry 5.0 面向可持续制造的人类感知调度：工业5.0时代动态作业车间调度研究综述

IF 11.4 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Robotics and Computer-integrated Manufacturing

Pub Date : 2025-09-25 DOI: 10.1016/j.rcim.2025.103143

Hudaifah Hudaifah , Haitham Saleh , Anas Alghazi , Ahmet Kolus , Umar Alturki , Sami Elferik

In the context of Industry 5.0, job scheduling must evolve beyond traditional efficiency-focused approaches to incorporate adaptability, sustainability, and human-centric approaches. Although Industry 4.0 technologies such as IoT, digital twins, and sensors have enabled real-time and dynamic-adaptive scheduling, most current systems still rely on static models and lack integrated consideration of environmental and human factors within dynamic scheduling contexts. To realize the vision of Industry 5.0 in practical applications, there is a growing need for dynamic scheduling methods that unify these dimensions. Given the limited research in this area, the present study proposes a comprehensive research framework for sustainable dynamic job scheduling, supported by structured conceptual models that explicitly outline how dynamic factors, environmental aspects, and human factors can be systematically incorporated into job scheduling problems. A systematic review of the literature is also conducted to assess recent progress and identify underexplored areas. The resulting framework is intended to provide a clear and structured foundation for future research aimed at developing intelligent, adaptive, eco-friendly, and human-aware scheduling systems aligned with the demands of Industry 5.0.

在工业5.0的背景下，作业调度必须超越传统的以效率为中心的方法，结合适应性、可持续性和以人为中心的方法。尽管物联网、数字孪生和传感器等工业4.0技术已经实现了实时和动态自适应调度，但目前大多数系统仍然依赖于静态模型，缺乏对动态调度环境和人为因素的综合考虑。为了在实际应用中实现工业5.0的愿景，越来越需要统一这些维度的动态调度方法。鉴于这一领域的研究有限，本研究提出了一个可持续动态作业调度的综合研究框架，并以结构化的概念模型为支持，该模型明确概述了动态因素、环境因素和人为因素如何系统地纳入作业调度问题。对文献进行了系统的回顾，以评估最近的进展并确定未开发的领域。由此产生的框架旨在为未来的研究提供一个清晰和结构化的基础，旨在开发符合工业5.0需求的智能、自适应、生态友好和人类感知的调度系统。

{"title":"Human-aware scheduling for sustainable manufacturing: A review of dynamic job shop scheduling in the era of Industry 5.0","authors":"Hudaifah Hudaifah , Haitham Saleh , Anas Alghazi , Ahmet Kolus , Umar Alturki , Sami Elferik","doi":"10.1016/j.rcim.2025.103143","DOIUrl":"10.1016/j.rcim.2025.103143","url":null,"abstract":"<div><div>In the context of Industry 5.0, job scheduling must evolve beyond traditional efficiency-focused approaches to incorporate adaptability, sustainability, and human-centric approaches. Although Industry 4.0 technologies such as IoT, digital twins, and sensors have enabled real-time and dynamic-adaptive scheduling, most current systems still rely on static models and lack integrated consideration of environmental and human factors within dynamic scheduling contexts. To realize the vision of Industry 5.0 in practical applications, there is a growing need for dynamic scheduling methods that unify these dimensions. Given the limited research in this area, the present study proposes a comprehensive research framework for sustainable dynamic job scheduling, supported by structured conceptual models that explicitly outline how dynamic factors, environmental aspects, and human factors can be systematically incorporated into job scheduling problems. A systematic review of the literature is also conducted to assess recent progress and identify underexplored areas. The resulting framework is intended to provide a clear and structured foundation for future research aimed at developing intelligent, adaptive, eco-friendly, and human-aware scheduling systems aligned with the demands of Industry 5.0.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"98 ","pages":"Article 103143"},"PeriodicalIF":11.4,"publicationDate":"2025-09-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145158951","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A vision-based self-calibration method for industrial robots using variable pose constraints 一种基于视觉的工业机器人变位姿自标定方法

IF 11.4 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Robotics and Computer-integrated Manufacturing

Pub Date : 2025-09-23 DOI: 10.1016/j.rcim.2025.103142

Yiyang Feng, Jianhui He, Jingbo Luo, Zaojun Fang, Chi Zhang, Guilin Yang

Among various geometrical constraints employed for robot self-calibration, the pose constraint by simultaneously restricting the position and orientation of the robot end-effector is the most comprehensive and effective constraint. However, as it is difficult to control the robot to precisely satisfy the pose constraints, a vision-based robot pose measurement system is designed, which mainly consists of two monochrome cameras fixed onto an adjustment stage and a pose target module mounted on the robot end-effector. Variable pose constraints are established when two or more robot poses are measured with two monochrome cameras at a fixed location. Based on the product-of-exponential (POE) formula, a new self-calibration model is formulated for industrial robots using variable pose constraint in which the robot pose errors are expressed in its tool frame and the position errors are decoupled from the orientation measurement errors. Therefore, the proposed self-calibration model is more accurate and robust than the conventional calibration model, in which the robot pose errors are expressed in its base frame and the position errors are coupled with the orientation measurement errors. Both simulations and experiments are conducted to validate the effectiveness of the proposed self-calibration method. Experimental results on the Aubo i5 robot demonstrate that after calibration, the average position error is reduced from 2.47 mm to 0.77 mm, and the average orientation error is reduced from 0.016 rad to 0.0039 rad.

在用于机器人自标定的各种几何约束中，同时约束机器人末端执行器位置和姿态的位姿约束是最全面、最有效的约束。然而，由于难以控制机器人精确满足姿态约束，设计了一种基于视觉的机器人姿态测量系统，该系统主要由固定在调节台上的两个单色摄像机和安装在机器人末端执行器上的姿态目标模块组成。当在固定位置用两台单色相机测量两个或多个机器人的姿态时，建立可变姿态约束。基于指数积（POE）公式，建立了一种基于变位姿约束的工业机器人自标定模型，该模型将机器人位姿误差表示在其刀架中，并将位置误差与姿态测量误差解耦。因此，所提出的自校准模型比传统的机器人位姿误差在基架中表示、位置误差与姿态测量误差耦合的自校准模型更准确、鲁棒。仿真和实验验证了所提出的自标定方法的有效性。在Aubo i5机器人上的实验结果表明，标定后的平均位置误差从2.47 mm减小到0.77 mm，平均方位误差从0.016 rad减小到0.0039 rad。

{"title":"A vision-based self-calibration method for industrial robots using variable pose constraints","authors":"Yiyang Feng, Jianhui He, Jingbo Luo, Zaojun Fang, Chi Zhang, Guilin Yang","doi":"10.1016/j.rcim.2025.103142","DOIUrl":"10.1016/j.rcim.2025.103142","url":null,"abstract":"<div><div>Among various geometrical constraints employed for robot self-calibration, the pose constraint by simultaneously restricting the position and orientation of the robot end-effector is the most comprehensive and effective constraint. However, as it is difficult to control the robot to precisely satisfy the pose constraints, a vision-based robot pose measurement system is designed, which mainly consists of two monochrome cameras fixed onto an adjustment stage and a pose target module mounted on the robot end-effector. Variable pose constraints are established when two or more robot poses are measured with two monochrome cameras at a fixed location. Based on the product-of-exponential (POE) formula, a new self-calibration model is formulated for industrial robots using variable pose constraint in which the robot pose errors are expressed in its tool frame and the position errors are decoupled from the orientation measurement errors. Therefore, the proposed self-calibration model is more accurate and robust than the conventional calibration model, in which the robot pose errors are expressed in its base frame and the position errors are coupled with the orientation measurement errors. Both simulations and experiments are conducted to validate the effectiveness of the proposed self-calibration method. Experimental results on the Aubo i5 robot demonstrate that after calibration, the average position error is reduced from 2.47 mm to 0.77 mm, and the average orientation error is reduced from 0.016 rad to 0.0039 rad.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"98 ","pages":"Article 103142"},"PeriodicalIF":11.4,"publicationDate":"2025-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145118871","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A generalised system for multi-mobile robot cooperation in smart manufacturing 智能制造中多移动机器人协作的通用系统

IF 11.4 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Robotics and Computer-integrated Manufacturing

Pub Date : 2025-09-23 DOI: 10.1016/j.rcim.2025.103139

Tianwei Zhang , Ning Wang , Yiming Yang , Ziya Wang

Since the advent of Industry 4.0, mobile collaborative robot technology has developed rapidly. However, several challenges still hinder the industrial application of mobile collaborative robots. These challenges include human–robot collaboration safety, the complexity of non-standard mobile collaborative task solutions, and the deployment timeliness of heterogeneous robots and mechanical structures. To address these challenges, this paper proposes a general and robust “cloud–edge-terminal-network-intelligence” multi-robot mobile collaboration system. This framework achieves efficient customised solutions by standardising and simplifying robot hardware and software integration. The solution focuses on three key issues: modular design, cloud–edge architecture in advanced manufacturing, and rapid deployment of heterogeneous multi-robots. The paper discusses robot safety and collaborative control issues and provides corresponding technical solutions.

自工业4.0时代到来以来，移动协同机器人技术得到了迅速发展。然而，一些挑战仍然阻碍着移动协作机器人的工业应用。这些挑战包括人机协作的安全性、非标准移动协作任务解决方案的复杂性以及异构机器人和机械结构的部署及时性。针对这些挑战，本文提出了一种通用、鲁棒的“云-边缘-终端-网络智能”多机器人移动协作系统。该框架通过标准化和简化机器人硬件和软件集成来实现高效的定制解决方案。该解决方案侧重于三个关键问题：模块化设计、先进制造中的云边缘架构和异构多机器人的快速部署。讨论了机器人的安全和协同控制问题，并提出了相应的技术解决方案。

引用次数: 0

Integrated scheduling of green production with flexible preventive maintenance and customer-side workers by a learning-based coevolutionary algorithm 基于学习的协同进化算法的柔性预防性维护和客户端工人的绿色生产集成调度

IF 11.4 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Robotics and Computer-integrated Manufacturing

Pub Date : 2025-09-19 DOI: 10.1016/j.rcim.2025.103140

Jingxing Zhang , Qianwang Deng , Qiang Luo , Kaidan Deng , Mengqi Liao , Yong Lei

Previous production scheduling studies on integrating preventive maintenance (PM) plans have overlooked the impact of arranging customer-side workers on the coupling of spare part delivery times, potentially leading to inefficient solutions. To address this gap, this study expands an integrated scheduling model of green two-stage hybrid flowshop production with flexible PM mode and customer-side workers for spare part replacement services. The model arranges tasks for the customer-side workers based on the relationship between delivery timetables, worker selection, replacement sequences, and equipment due time windows, aiming to maximize the total customer satisfaction. Another objective is to minimize the total energy consumption during production, maintenance and idle processes. To solve the large-scale instances, a double deep Q-network-based coevolutionary algorithm (shorten to DDQCA) is proposed, incorporating a six-layer chromosome encoding scheme. In DDQCA, double deep Q-networks are trained online to guide the selection of crossover methods for coevolution. Additionally, the DDQCA incorporates a hybrid initialization operator, two objectives-oriented local search methods and a proposition-based PM strategy to enhance search performance. Finally, comprehensive experiments are conducted to validate the effectiveness of the algorithm. In addition, the results also demonstrate that the proposed integrated scheduling with flexible PM mode can improve energy efficiency but significantly customer satisfaction compared to the classical mode.

之前关于集成预防性维护（PM）计划的生产调度研究忽略了安排客户端工人对备件交付时间耦合的影响，可能导致低效的解决方案。为了解决这一差距，本研究扩展了一个绿色两阶段混合流水车间生产的集成调度模型，该模型具有灵活的PM模式和备件更换服务的客户端工人。该模型根据交货时间表、工人选择、更换顺序和设备到期时间窗口之间的关系来安排客户端工人的任务，以最大限度地提高客户的总满意度。另一个目标是尽量减少生产、维护和闲置过程中的总能耗。为了解决大规模实例，提出了一种基于双深度q网络的协同进化算法（简称DDQCA），该算法采用六层染色体编码方案。在DDQCA中，在线训练双深度q网络来指导协同进化交叉方法的选择。此外，DDQCA还结合了一个混合初始化算子、两种面向目标的局部搜索方法和一个基于命题的PM策略来提高搜索性能。最后进行了全面的实验，验证了算法的有效性。此外，结果还表明，与传统模式相比，柔性PM模式下的集成调度可以提高能源效率，但客户满意度显著提高。

{"title":"Integrated scheduling of green production with flexible preventive maintenance and customer-side workers by a learning-based coevolutionary algorithm","authors":"Jingxing Zhang , Qianwang Deng , Qiang Luo , Kaidan Deng , Mengqi Liao , Yong Lei","doi":"10.1016/j.rcim.2025.103140","DOIUrl":"10.1016/j.rcim.2025.103140","url":null,"abstract":"<div><div>Previous production scheduling studies on integrating preventive maintenance (PM) plans have overlooked the impact of arranging customer-side workers on the coupling of spare part delivery times, potentially leading to inefficient solutions. To address this gap, this study expands an integrated scheduling model of green two-stage hybrid flowshop production with flexible PM mode and customer-side workers for spare part replacement services. The model arranges tasks for the customer-side workers based on the relationship between delivery timetables, worker selection, replacement sequences, and equipment due time windows, aiming to maximize the total customer satisfaction. Another objective is to minimize the total energy consumption during production, maintenance and idle processes. To solve the large-scale instances, a double deep Q-network-based coevolutionary algorithm (shorten to DDQCA) is proposed, incorporating a six-layer chromosome encoding scheme. In DDQCA, double deep Q-networks are trained online to guide the selection of crossover methods for coevolution. Additionally, the DDQCA incorporates a hybrid initialization operator, two objectives-oriented local search methods and a proposition-based PM strategy to enhance search performance. Finally, comprehensive experiments are conducted to validate the effectiveness of the algorithm. In addition, the results also demonstrate that the proposed integrated scheduling with flexible PM mode can improve energy efficiency but significantly customer satisfaction compared to the classical mode.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"98 ","pages":"Article 103140"},"PeriodicalIF":11.4,"publicationDate":"2025-09-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145093755","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0