Robotics and Computer-integrated Manufacturing最新文献_第3页

Toward generalizable robotic assembly: A prior-guided deep reinforcement learning approach with multi-sensor information 面向一般化机器人装配：基于多传感器信息的先验引导深度强化学习方法

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

Robotics and Computer-integrated Manufacturing

Pub Date : 2026-01-22 DOI: 10.1016/j.rcim.2026.103242

Zilu Zhu , Yongkui Liu , Qianji Wang , Zinan Wang , Lihui Wang , Sichao Liu , Bin Zi , Lin Zhang

The rise of personalized manufacturing presents significant challenges for robotic assembly. While learning-based methods offer promising solutions, they often suffer from low training efficiency and poor generalization. To address these limitations, this paper proposes an efficient prior-guided (PG) deep reinforcement learning (DRL) approach for generalizable robotic assembly using multi-sensor information. First, a phased multi-sensor information fusion method is introduced. Then, a visual feature extraction method that combines MobileNetV3-Lite with conventional digital image processing and a rule-based force feature extraction method are designed to extract lower-dimensional features as prior-guided knowledge. Based on the methods above, a Soft Actor-Critic (SAC) algorithm that integrates Gated Recurrent Unit (GRU) network architecture with PG is proposed, thereby enabling efficient assembly skill learning. Simulations and physical experiments with respect to three typical assembly skills, i.e., search, alignment, and insertion, are conducted. Results indicate that, compared with the baseline SAC algorithm, our feature extraction method reduces visual feature dimensions by 93.75% and provides accurate prior-guided knowledge for DRL. The proposed assembly skill learning algorithm achieves a 30.16% reduction in average training time and a 16.82% decrease in average completion step. Furthermore, all learned skills can be rapidly transferred across different objects, and all assembly tasks are completed efficiently and compliantly with an average success rate of 96.86%.

个性化制造的兴起对机器人装配提出了重大挑战。虽然基于学习的方法提供了很好的解决方案，但它们往往存在训练效率低和泛化能力差的问题。为了解决这些限制，本文提出了一种有效的先验引导（PG）深度强化学习（DRL）方法，用于使用多传感器信息的可泛化机器人装配。首先，介绍了一种分阶段的多传感器信息融合方法。然后，设计了MobileNetV3-Lite与传统数字图像处理相结合的视觉特征提取方法和基于规则的力特征提取方法，作为先验引导知识提取低维特征。在此基础上，提出了一种将门控循环单元（GRU）网络体系结构与PG相结合的软Actor-Critic （SAC）算法，从而实现了高效的装配技能学习。针对三种典型装配技能，即搜索、对准和插入，进行了仿真和物理实验。结果表明，与基线SAC算法相比，我们的特征提取方法将视觉特征维数降低了93.75%，为DRL提供了准确的先验引导知识。提出的装配技能学习算法平均训练时间减少30.16%，平均完成步长减少16.82%。此外，所有学习到的技能都可以在不同的对象之间快速转移，所有装配任务都能高效、合规地完成，平均成功率为96.86%。

{"title":"Toward generalizable robotic assembly: A prior-guided deep reinforcement learning approach with multi-sensor information","authors":"Zilu Zhu , Yongkui Liu , Qianji Wang , Zinan Wang , Lihui Wang , Sichao Liu , Bin Zi , Lin Zhang","doi":"10.1016/j.rcim.2026.103242","DOIUrl":"10.1016/j.rcim.2026.103242","url":null,"abstract":"<div><div>The rise of personalized manufacturing presents significant challenges for robotic assembly. While learning-based methods offer promising solutions, they often suffer from low training efficiency and poor generalization. To address these limitations, this paper proposes an efficient prior-guided (PG) deep reinforcement learning (DRL) approach for generalizable robotic assembly using multi-sensor information. First, a phased multi-sensor information fusion method is introduced. Then, a visual feature extraction method that combines MobileNetV3-Lite with conventional digital image processing and a rule-based force feature extraction method are designed to extract lower-dimensional features as prior-guided knowledge. Based on the methods above, a Soft Actor-Critic (SAC) algorithm that integrates Gated Recurrent Unit (GRU) network architecture with PG is proposed, thereby enabling efficient assembly skill learning. Simulations and physical experiments with respect to three typical assembly skills, i.e., search, alignment, and insertion, are conducted. Results indicate that, compared with the baseline SAC algorithm, our feature extraction method reduces visual feature dimensions by 93.75% and provides accurate prior-guided knowledge for DRL. The proposed assembly skill learning algorithm achieves a 30.16% reduction in average training time and a 16.82% decrease in average completion step. Furthermore, all learned skills can be rapidly transferred across different objects, and all assembly tasks are completed efficiently and compliantly with an average success rate of 96.86%.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"100 ","pages":"Article 103242"},"PeriodicalIF":11.4,"publicationDate":"2026-01-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146033276","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 digital-twin framework for assembling of cylindrical parts 用于圆柱形零件装配的数字孪生框架

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

Robotics and Computer-integrated Manufacturing

Pub Date : 2026-01-21 DOI: 10.1016/j.rcim.2026.103240

Yimin Song , Chen Li , Binbin Lian , Qi Li , Tao Sun

Digital twin (DT) has been recognized as a promising technology for enhanced planning, monitoring and control of automatic assembly, with the capability of efficiency, adaptability and flexibility. While most of DT-based assembly focus on electronic products, limited attention has been paid to the assembling of heavy and large-size product with tight tolerance. This paper presents a DT model facilitating prediction and real-time adjustment for intelligent assembling of cylindrical parts. Vision guided feature fitting and coordinate frame construction are presented. Herein, assembly targets are defined considering eight sets of pin and hole, and the contacting planes. This approach improves the assembly success rate. To ensure efficient and robust robot assembly, we proposed a prediction model based on the DT system. Unknown errors and uncertainty of physical space are considered by small displacement torsor (SDT) theory and Monte Carlo simulation (MCS). Assembly planning and execution would efficiently adjust guided by the prediction result. A middle point is set in the robot planning that leaves pure translation in the docking phase. Real-time adjustment method is proposed to accurately assemble the cylindrical parts. Simulations and experiments are carried out to verify the effectiveness and feasibility of the proposed DT-based assembling method. The results show that the prediction results are the same as the actual assembly. Our assembly strategy achieves 97.31% success rate. By employing the assembly strategy and real-time adjustment, our method ensures that the majority of axes mismatch is below 0.1mm/0.05 deg, plane non-contacting below 0.05 mm.

数字孪生技术（Digital twin， DT）以其高效、适应性强、灵活的特点，被认为是一种很有前途的自动化装配规划、监测和控制技术。目前，基于3d打印技术的装配大多集中在电子产品上，而对重、大尺寸、公差要求高的产品的装配关注甚少。针对圆柱件智能装配，提出了一种便于预测和实时调整的DT模型。提出了视觉引导特征拟合和坐标框架构建方法。其中，考虑8组销孔和接触面，定义装配目标。这种方法提高了装配成功率。为了保证机器人装配的效率和鲁棒性，提出了一种基于DT系统的预测模型。小位移变形量（SDT）理论和蒙特卡罗模拟（MCS）考虑了物理空间的未知误差和不确定性。在预测结果的指导下，有效地调整装配规划和执行。在机器人规划中设置一个中间点，在对接阶段保留纯平移。提出了圆柱零件精确装配的实时调整方法。仿真和实验验证了该方法的有效性和可行性。结果表明，预测结果与实际装配相吻合。我们的装配策略达到97.31%的成功率。通过采用装配策略和实时调整，保证了大部分轴错配在0.1mm/0.05°以下，平面不接触在0.05 mm以下。

{"title":"A digital-twin framework for assembling of cylindrical parts","authors":"Yimin Song , Chen Li , Binbin Lian , Qi Li , Tao Sun","doi":"10.1016/j.rcim.2026.103240","DOIUrl":"10.1016/j.rcim.2026.103240","url":null,"abstract":"<div><div>Digital twin (DT) has been recognized as a promising technology for enhanced planning, monitoring and control of automatic assembly, with the capability of efficiency, adaptability and flexibility. While most of DT-based assembly focus on electronic products, limited attention has been paid to the assembling of heavy and large-size product with tight tolerance. This paper presents a DT model facilitating prediction and real-time adjustment for intelligent assembling of cylindrical parts. Vision guided feature fitting and coordinate frame construction are presented. Herein, assembly targets are defined considering eight sets of pin and hole, and the contacting planes. This approach improves the assembly success rate. To ensure efficient and robust robot assembly, we proposed a prediction model based on the DT system. Unknown errors and uncertainty of physical space are considered by small displacement torsor (SDT) theory and Monte Carlo simulation (MCS). Assembly planning and execution would efficiently adjust guided by the prediction result. A middle point is set in the robot planning that leaves pure translation in the docking phase. Real-time adjustment method is proposed to accurately assemble the cylindrical parts. Simulations and experiments are carried out to verify the effectiveness and feasibility of the proposed DT-based assembling method. The results show that the prediction results are the same as the actual assembly. Our assembly strategy achieves 97.31% success rate. By employing the assembly strategy and real-time adjustment, our method ensures that the majority of axes mismatch is below 0.1mm/0.05 deg, plane non-contacting below 0.05 mm.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"100 ","pages":"Article 103240"},"PeriodicalIF":11.4,"publicationDate":"2026-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146014798","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

Physics-informed prediction of modal parameters and stability analysis for robotic mirror milling 机器人铣镜模态参数的物理预测与稳定性分析

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

Robotics and Computer-integrated Manufacturing

Pub Date : 2026-01-20 DOI: 10.1016/j.rcim.2026.103237

Kun Chen , Haonan Ma , Chenghao Huang , Sheng Xu , Peng Xu , Bing Li

Mirror milling technology is widely used in the aerospace industry for manufacturing thin-walled parts, yet existing machine tool-based mirror milling systems are costly and inflexible. Robotic mirror milling system is a cost-effective and flexible alternative to machine tools. However, the modal parameters of the mirror-arranged robots vary with their postures, and the robots’ low stiffness, coupled with the flexibility of thin-walled parts, leads to unstable milling processes. To address these challenges, a physics-informed framework is proposed for modal parameters measurement, prediction, and optimization, thereby analyzing the robotic mirror milling stability. First, the robot’s vibration characteristics are examined through transfer matrices of dynamic models, while the robot’s modal parameters are collected at uniform configurations in joint space. Using these characteristics and measurements as physical constraints and training sets, a modified multi-task Gaussian process regression is developed to predict the modal parameters, with the results further optimized through the Bayesian derivation. This two-step process forms the physics-informed modal parameters prediction method. Then, the obtained modal parameters are utilized to construct the robotic mirror milling system’s dynamic model, which can analyze its milling stability. Simulations and experiments are conducted to confirm these theories and algorithms.

镜面铣削技术在航空航天工业中广泛应用于制造薄壁零件，但现有的基于机床的镜面铣削系统成本高且不灵活。机器人镜面铣削系统是一种具有成本效益和灵活性的机床替代方案。然而，镜面排列机器人的模态参数随其姿态而变化，并且机器人的低刚度加上薄壁零件的柔韧性导致铣削过程不稳定。为了解决这些挑战，提出了一个物理信息框架，用于模态参数的测量、预测和优化，从而分析机器人镜面铣削稳定性。首先，通过动力学模型的传递矩阵分析了机器人的振动特性，同时在关节空间中采集了机器人的均匀构型模态参数。利用这些特征和测量作为物理约束和训练集，提出了一种改进的多任务高斯过程回归来预测模态参数，并通过贝叶斯推导进一步优化结果。这两步过程形成了基于物理的模态参数预测方法。然后，利用得到的模态参数构建机器人镜面铣削系统的动力学模型，分析其铣削稳定性。通过仿真和实验验证了这些理论和算法。

{"title":"Physics-informed prediction of modal parameters and stability analysis for robotic mirror milling","authors":"Kun Chen , Haonan Ma , Chenghao Huang , Sheng Xu , Peng Xu , Bing Li","doi":"10.1016/j.rcim.2026.103237","DOIUrl":"10.1016/j.rcim.2026.103237","url":null,"abstract":"<div><div>Mirror milling technology is widely used in the aerospace industry for manufacturing thin-walled parts, yet existing machine tool-based mirror milling systems are costly and inflexible. Robotic mirror milling system is a cost-effective and flexible alternative to machine tools. However, the modal parameters of the mirror-arranged robots vary with their postures, and the robots’ low stiffness, coupled with the flexibility of thin-walled parts, leads to unstable milling processes. To address these challenges, a physics-informed framework is proposed for modal parameters measurement, prediction, and optimization, thereby analyzing the robotic mirror milling stability. First, the robot’s vibration characteristics are examined through transfer matrices of dynamic models, while the robot’s modal parameters are collected at uniform configurations in joint space. Using these characteristics and measurements as physical constraints and training sets, a modified multi-task Gaussian process regression is developed to predict the modal parameters, with the results further optimized through the Bayesian derivation. This two-step process forms the physics-informed modal parameters prediction method. Then, the obtained modal parameters are utilized to construct the robotic mirror milling system’s dynamic model, which can analyze its milling stability. Simulations and experiments are conducted to confirm these theories and algorithms.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"100 ","pages":"Article 103237"},"PeriodicalIF":11.4,"publicationDate":"2026-01-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146014514","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 negative-pressure-based surface-compliant constant-force grinding end-effector for climbing machining robots 一种基于负压的表面柔性恒力磨削端部执行器

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

Robotics and Computer-integrated Manufacturing

Pub Date : 2026-01-17 DOI: 10.1016/j.rcim.2026.103241

Zeyu Gong , Junhui Huang , Ying Shi , Yaonan Wang , Bo Tao

Climbing robots, with their unique adsorption and locomotion mechanisms, overcome the accessibility limitations of traditional robots, offering innovative solutions for manufacturing large and complex components such as aircraft skins and wind turbine blades. This paper proposes a specialized grinding end-effector with surface-conforming and force control capabilities for climbing robots, targeting surface finishing tasks. A surface-adaptive module driven by vacuum suction force is introduced, enabling both 3-degree-of-freedom posture self-adjustment and dust collection for the end-effector using only a single centrifugal fan. Building upon this, a force control module decoupled from posture adaptation is designed. Utilizing voice coil motor (VCM) actuation, it integrates parallel force transmission and feedback mechanisms among the VCM, grinding head, and force sensor, combined with PID force control algorithms, to achieve high-precision grinding force regulation. Experiments conducted on both curved aluminum alloy skins and real aircraft skin workpiece demonstrate that the proposed end-effector achieves high-accuracy adaptive normal-direction tracking and constant grinding force control, with posture errors as low as 1.59° and average force errors of 0.18 N. The grinding process effectively activates workpiece surfaces and significantly improves surface roughness uniformity, proving that the proposed end-effector empowers climbing robots to perform high-quality surface finishing operations on curved structures.

攀爬机器人以其独特的吸附和运动机制，克服了传统机器人的可及性限制，为制造飞机蒙皮和风力涡轮机叶片等大型复杂部件提供了创新的解决方案。本文提出了一种具有表面一致性和力控制能力的爬行机器人专用磨削末端执行器，用于表面精加工任务。介绍了一种由真空吸力驱动的表面自适应模块，只需一个离心风机即可实现末端执行器的3自由度姿态自调节和除尘。在此基础上，设计了与姿态自适应解耦的力控制模块。利用音圈电机（VCM）驱动，集成了VCM、磨头、力传感器之间的并联力传递和反馈机构，结合PID力控制算法，实现磨削力的高精度调节。在曲面铝合金蒙皮和实际飞机蒙皮工件上的实验表明，该末端执行器实现了高精度自适应法向跟踪和恒磨削力控制，姿态误差低至1.59°，平均力误差为0.18 n，磨削过程有效激活了工件表面，显著提高了表面粗糙度均匀性。证明所提出的末端执行器使攀爬机器人能够在弯曲结构上执行高质量的表面加工操作。

{"title":"A negative-pressure-based surface-compliant constant-force grinding end-effector for climbing machining robots","authors":"Zeyu Gong , Junhui Huang , Ying Shi , Yaonan Wang , Bo Tao","doi":"10.1016/j.rcim.2026.103241","DOIUrl":"10.1016/j.rcim.2026.103241","url":null,"abstract":"<div><div>Climbing robots, with their unique adsorption and locomotion mechanisms, overcome the accessibility limitations of traditional robots, offering innovative solutions for manufacturing large and complex components such as aircraft skins and wind turbine blades. This paper proposes a specialized grinding end-effector with surface-conforming and force control capabilities for climbing robots, targeting surface finishing tasks. A surface-adaptive module driven by vacuum suction force is introduced, enabling both 3-degree-of-freedom posture self-adjustment and dust collection for the end-effector using only a single centrifugal fan. Building upon this, a force control module decoupled from posture adaptation is designed. Utilizing voice coil motor (VCM) actuation, it integrates parallel force transmission and feedback mechanisms among the VCM, grinding head, and force sensor, combined with PID force control algorithms, to achieve high-precision grinding force regulation. Experiments conducted on both curved aluminum alloy skins and real aircraft skin workpiece demonstrate that the proposed end-effector achieves high-accuracy adaptive normal-direction tracking and constant grinding force control, with posture errors as low as 1.59° and average force errors of 0.18 N. The grinding process effectively activates workpiece surfaces and significantly improves surface roughness uniformity, proving that the proposed end-effector empowers climbing robots to perform high-quality surface finishing operations on curved structures.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"100 ","pages":"Article 103241"},"PeriodicalIF":11.4,"publicationDate":"2026-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145979602","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

Research on a curvature- and time-optimal corner smoothing method for machining paths of a 6-axis hybrid machining robot 六轴混合加工机器人加工路径的曲率和时间最优角点平滑方法研究

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

Robotics and Computer-integrated Manufacturing

Pub Date : 2026-01-17 DOI: 10.1016/j.rcim.2026.103239

Yue Ma , Honggui Peng , Haitao Liu , Bin Li , Qi Liu , Long Chen , Dun Peng

Discontinuous corner smoothing in machining paths and excessive rapid traverse paths degrade both the machining quality and efficiency of hybrid robots. Therefore, two algorithms are proposed for dealing with the trajectory planning of a 6-axis hybrid machining robot within a single machining zone and across two machining zones, respectively. The constraints on the curvatures and position errors of smoothing path segments are considered in corner smoothing planning within a single machining zone, while the constraints on the curvatures and obstacle geometries are employed in planning obstacle-avoidance paths across two machining zones. Both of the above optimization problems are solved by adjusting the weights of spline curves through the Quantum-behaved particle swarm optimization (QPSO) algorithm combined with the greedy algorithm. Furthermore, the time allocation is optimized using the QPSO algorithm combined with the moving window planning method to enhance the machining efficiency of toolpaths. The comparative simulation results within a single zone indicate that, under identical position error constraints, the proposed method achieves a maximum curvature reduction of 27.29 % relative to the B-spline method, and reduces computational time by 48.35 % and 53.31 % compared to the PSO algorithm and the genetic algorithm, respectively. The simulation results across two machining zones demonstrate that the proposed method is capable of generating curvature-optimal obstacle-avoidance toolpaths for obstacles with varying geometries. Additionally, the simulation results of butterfly and maple leaf contours show that the proposed method reduces machining time by 52.59 % and 40.92 % compared to the Bezier curve method and the Clothoid spline method, respectively. The experimental results show that the predicted curvatures are in close agreement with the measured ones along the butterfly and maple leaf contours, with a maximum error of 2.99 %. Furthermore, the measured tracking errors of the actuated joints are maintained within ±0.028 mm and ±1.3 × 10⁻⁴ rad for the parallel mechanism and serial wrist, respectively. These results fully demonstrate the effectiveness of the proposed method in enhancing machining efficiency and motion smoothness in the planning of curvature-optimal corner smoothing toolpaths.

加工路径的不连续角点平滑和过快的遍历路径会降低混合动力机器人的加工质量和效率。因此，针对六轴混合加工机器人在单加工区域内和跨加工区域的轨迹规划问题，提出了两种算法。在单个加工区域内进行角点平滑规划时，考虑了对平滑路径段曲率和位置误差的约束；在规划跨两个加工区域的避障路径时，考虑了曲率和障碍物几何形状的约束。采用量子粒子群优化（QPSO）算法结合贪心算法，通过调整样条曲线的权值来解决上述两个优化问题。在此基础上，利用QPSO算法结合运动窗规划方法优化了刀具轨迹的时间分配，提高了刀具轨迹的加工效率。单个区域内的对比仿真结果表明，在相同位置误差约束条件下，该方法相对于b样条法曲率最大减小27.29%，计算时间相对于粒子群算法和遗传算法分别减少48.35%和53.31%。跨两个加工区域的仿真结果表明，该方法能够针对不同几何形状的障碍物生成曲率最优的避障刀具路径。此外，蝴蝶和枫叶轮廓的仿真结果表明，该方法与Bezier曲线法和clocloid样条法相比，加工时间分别缩短了52.59%和40.92%。实验结果表明，所预测的沿蝴蝶和枫叶轮廓的曲率与实测值吻合较好，最大误差为2.99%。此外，并联机构和串联手腕的驱动关节的测量跟踪误差分别保持在±0.028 mm和±1.3 × 10⁻⁴rad。这些结果充分证明了该方法在曲率最优角点平滑刀具轨迹规划中提高加工效率和运动平稳性的有效性。

{"title":"Research on a curvature- and time-optimal corner smoothing method for machining paths of a 6-axis hybrid machining robot","authors":"Yue Ma , Honggui Peng , Haitao Liu , Bin Li , Qi Liu , Long Chen , Dun Peng","doi":"10.1016/j.rcim.2026.103239","DOIUrl":"10.1016/j.rcim.2026.103239","url":null,"abstract":"<div><div>Discontinuous corner smoothing in machining paths and excessive rapid traverse paths degrade both the machining quality and efficiency of hybrid robots. Therefore, two algorithms are proposed for dealing with the trajectory planning of a 6-axis hybrid machining robot within a single machining zone and across two machining zones, respectively. The constraints on the curvatures and position errors of smoothing path segments are considered in corner smoothing planning within a single machining zone, while the constraints on the curvatures and obstacle geometries are employed in planning obstacle-avoidance paths across two machining zones. Both of the above optimization problems are solved by adjusting the weights of spline curves through the Quantum-behaved particle swarm optimization (QPSO) algorithm combined with the greedy algorithm. Furthermore, the time allocation is optimized using the QPSO algorithm combined with the moving window planning method to enhance the machining efficiency of toolpaths. The comparative simulation results within a single zone indicate that, under identical position error constraints, the proposed method achieves a maximum curvature reduction of 27.29 % relative to the B-spline method, and reduces computational time by 48.35 % and 53.31 % compared to the PSO algorithm and the genetic algorithm, respectively. The simulation results across two machining zones demonstrate that the proposed method is capable of generating curvature-optimal obstacle-avoidance toolpaths for obstacles with varying geometries. Additionally, the simulation results of butterfly and maple leaf contours show that the proposed method reduces machining time by 52.59 % and 40.92 % compared to the Bezier curve method and the Clothoid spline method, respectively. The experimental results show that the predicted curvatures are in close agreement with the measured ones along the butterfly and maple leaf contours, with a maximum error of 2.99 %. Furthermore, the measured tracking errors of the actuated joints are maintained within ±0.028 mm and ±1.3 × 10⁻⁴ rad for the parallel mechanism and serial wrist, respectively. These results fully demonstrate the effectiveness of the proposed method in enhancing machining efficiency and motion smoothness in the planning of curvature-optimal corner smoothing toolpaths.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"100 ","pages":"Article 103239"},"PeriodicalIF":11.4,"publicationDate":"2026-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145995476","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

Towards efficient robotic milling: A novel time-varying feed rate control framework via system analysis and model parameters identification 面向高效机器人铣削：基于系统分析和模型参数辨识的新型时变进给速率控制框架

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

Robotics and Computer-integrated Manufacturing

Pub Date : 2026-01-16 DOI: 10.1016/j.rcim.2026.103229

Qingyu Peng , Wenlong Li , Hanyu Zhang , Wei Xu , Gang Wang , Han Ding

In robotic milling, feed rate is a pivotal factor in machining efficiency and quality. A common strategy is to maintain a lower feed rate to ensure system stability. However, this method lacks specific analysis of the robotic milling system, and the feed rate is typically set conservatively based on empirical experience. To address this limitation and achieve both high efficiency and machining quality, a novel time-varying feed rate control framework is proposed. The framework first schedules a time-varying feed rate through a system analysis of robotic milling, which accounts for the robot’s kinematic performance and the maximum vibration amplitude of thin-walled parts. It then introduces a parameter identification module that estimates the dynamic feed rate model parameters, thereby enhancing the designed controller and ensuring that the actual feed rate closely tracks the scheduled feed rate. Furthermore, an active controller with an event-triggered mechanism is proposed to update the identified parameters in real time, ensuring improved control accuracy. Experimental results validate the effectiveness of the proposed method, demonstrating that the machining efficiencies of two different skin parts with varying dimensions are improved by 29.3% and 41.5%, respectively, while maintaining machining quality. In addition, application verification experiment on a real horizontal tail skin part further confirms the practical potential of the proposed method.

在机器人铣削中，进给速度是影响加工效率和加工质量的关键因素。一种常见的策略是保持较低的进给速率以确保系统的稳定性。然而，该方法缺乏对机器人铣削系统的具体分析，并且通常根据经验保守地设置进给速率。为了解决这一限制，同时实现高效率和加工质量，提出了一种新的时变进给速度控制框架。该框架首先通过对机器人铣削的系统分析来调度一个时变进给速率，该进给速率考虑了机器人的运动性能和薄壁零件的最大振动幅值。然后引入参数识别模块来估计动态进给速率模型参数，从而增强设计的控制器，并确保实际进给速率与计划进给速率密切跟踪。在此基础上，提出了一种带有事件触发机制的主动控制器，实时更新识别参数，保证了控制精度的提高。实验结果验证了该方法的有效性，在保证加工质量的前提下，两种不同尺寸蒙皮零件的加工效率分别提高了29.3%和41.5%。另外，在实际水平尾皮上进行了应用验证实验，进一步验证了该方法的实用潜力。

{"title":"Towards efficient robotic milling: A novel time-varying feed rate control framework via system analysis and model parameters identification","authors":"Qingyu Peng , Wenlong Li , Hanyu Zhang , Wei Xu , Gang Wang , Han Ding","doi":"10.1016/j.rcim.2026.103229","DOIUrl":"10.1016/j.rcim.2026.103229","url":null,"abstract":"<div><div>In robotic milling, feed rate is a pivotal factor in machining efficiency and quality. A common strategy is to maintain a lower feed rate to ensure system stability. However, this method lacks specific analysis of the robotic milling system, and the feed rate is typically set conservatively based on empirical experience. To address this limitation and achieve both high efficiency and machining quality, a novel time-varying feed rate control framework is proposed. The framework first schedules a time-varying feed rate through a system analysis of robotic milling, which accounts for the robot’s kinematic performance and the maximum vibration amplitude of thin-walled parts. It then introduces a parameter identification module that estimates the dynamic feed rate model parameters, thereby enhancing the designed controller and ensuring that the actual feed rate closely tracks the scheduled feed rate. Furthermore, an active controller with an event-triggered mechanism is proposed to update the identified parameters in real time, ensuring improved control accuracy. Experimental results validate the effectiveness of the proposed method, demonstrating that the machining efficiencies of two different skin parts with varying dimensions are improved by 29.3% and 41.5%, respectively, while maintaining machining quality. In addition, application verification experiment on a real horizontal tail skin part further confirms the practical potential of the proposed method.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"100 ","pages":"Article 103229"},"PeriodicalIF":11.4,"publicationDate":"2026-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145979601","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

Large language models for high-level computer-aided process planning in a distributed manufacturing paradigm 分布式制造范式中用于高级计算机辅助工艺规划的大型语言模型

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

Robotics and Computer-integrated Manufacturing

Pub Date : 2026-01-12 DOI: 10.1016/j.rcim.2026.103233

Emmanuel Stathatos , Panorios Benardos , George-Christopher Vosniakos , Dennis Gross , Helge Spieker , Arnaud Gotlieb

This study applies Large Language Models (LLMs) to high-level Computer-Aided Process Planning (CAPP) in a distributed manufacturing context. It aims to generate alternative, feasible process chains for production of a wide range of parts. Parts are encoded in a custom encoding scheme supporting diverse part overall shapes, geometrical features within them, and corresponding manufacturing processes. The CAPP problem is formulated as a sequence prediction task, where a GPT-2-based LLM generates process chains autoregressively. To train and test the LLM a synthetic dataset of 7,840 unique parts and their alternative process chains was generated using expert-driven rule-based logic. The LLM is trained from scratch using a tokenization scheme treating part features and processes uniformly as discrete tokens, special tokens being employed to control sequence flow. Performance evaluation was performed for systematically reducing the size of the dataset. Finally, even with as little as 5% of the training data, the LLM achieves over 99% accuracy at the process chain-level. The extremely few spotted errors mainly involve minor secondary process mispredictions without critical failures. For comparison, a Recurrent Neural Network (RNN) was also trained with the same dataset. Since manufacturing data stemming from experts and not from sensors is notoriously difficult to collect, training a machine learning model with a dataset that is as small as possible is of utmost importance. In this light, the LLM proved superior to RNN, in fact emphatically so, the more the training dataset was limited.

本研究将大型语言模型（llm）应用于分布式制造环境中的高级计算机辅助工艺规划（CAPP）。它的目的是产生替代的，可行的工艺链生产范围广泛的零件。零件采用自定义编码方案进行编码，支持不同的零件总体形状、几何特征和相应的制造工艺。CAPP问题被表述为序列预测任务，其中基于gpt -2的LLM自回归生成过程链。为了训练和测试LLM，使用专家驱动的基于规则的逻辑生成了7,840个独特零件及其替代工艺链的合成数据集。LLM使用标记化方案从头开始训练，将部分特征和过程统一处理为离散标记，使用特殊标记来控制序列流。对系统地减少数据集的大小进行了性能评估。最后，即使只有5%的训练数据，LLM在过程链级别上也达到了99%以上的准确率。极少发现的错误主要涉及次要过程的错误预测，而没有出现严重故障。为了进行比较，我们还使用相同的数据集训练了一个递归神经网络（RNN）。由于来自专家而非传感器的制造数据很难收集，因此使用尽可能小的数据集训练机器学习模型至关重要。从这个角度来看，LLM证明优于RNN，事实上，训练数据集越有限，就越明显优于RNN。

{"title":"Large language models for high-level computer-aided process planning in a distributed manufacturing paradigm","authors":"Emmanuel Stathatos , Panorios Benardos , George-Christopher Vosniakos , Dennis Gross , Helge Spieker , Arnaud Gotlieb","doi":"10.1016/j.rcim.2026.103233","DOIUrl":"10.1016/j.rcim.2026.103233","url":null,"abstract":"<div><div>This study applies Large Language Models (LLMs) to high-level Computer-Aided Process Planning (CAPP) in a distributed manufacturing context. It aims to generate alternative, feasible process chains for production of a wide range of parts. Parts are encoded in a custom encoding scheme supporting diverse part overall shapes, geometrical features within them, and corresponding manufacturing processes. The CAPP problem is formulated as a sequence prediction task, where a GPT-2-based LLM generates process chains autoregressively. To train and test the LLM a synthetic dataset of 7,840 unique parts and their alternative process chains was generated using expert-driven rule-based logic. The LLM is trained from scratch using a tokenization scheme treating part features and processes uniformly as discrete tokens, special tokens being employed to control sequence flow. Performance evaluation was performed for systematically reducing the size of the dataset. Finally, even with as little as 5% of the training data, the LLM achieves over 99% accuracy at the process chain-level. The extremely few spotted errors mainly involve minor secondary process mispredictions without critical failures. For comparison, a Recurrent Neural Network (RNN) was also trained with the same dataset. Since manufacturing data stemming from experts and not from sensors is notoriously difficult to collect, training a machine learning model with a dataset that is as small as possible is of utmost importance. In this light, the LLM proved superior to RNN, in fact emphatically so, the more the training dataset was limited.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"100 ","pages":"Article 103233"},"PeriodicalIF":11.4,"publicationDate":"2026-01-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145956532","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

Improve geometric accuracy in robotic forming 提高机器人成形的几何精度

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

Robotics and Computer-integrated Manufacturing

Pub Date : 2026-01-12 DOI: 10.1016/j.rcim.2026.103236

Yanrong Zhang , Zeran Hou , Fan Yang , Bernd Kuhlenkötter , Antonio Sánchez Egea , Junying Min

Robotic forming is a die-less manufacturing process that utilizes single or multiple industrial robots to incrementally deform metal sheets into complex, customized parts. Owing to its high flexibility, it has become an important research direction in the field of intelligent manufacturing due to its high flexibility. However, issues such as complex local deformation mechanisms and insufficient support during robotic forming often lead to low geometric accuracy in fabricated parts, which is one of the most critical factors hindering the broader application of robotic forming in high-precision manufacturing. In this context, this paper presents a comprehensive review of the methods for improving geometric accuracy in robotic forming. Firstly, to address the core issue of geometric accuracy, the categories and origins of geometric errors in robotic forming are systematically examined. Subsequently, typical precision control strategies and related research are categorized into three aspects: process innovation, optimization of process parameters, and tool path planning and compensation. Current challenges associated with each strategy are also summarized. Finally, potential future research directions are discussed, incorporating advanced technologies such as multi-robot forming, artificial intelligence, multi-source data fusion, and digital twins.

机器人成形是一种无模具制造工艺，利用单个或多个工业机器人将金属板材逐渐变形成复杂的定制零件。由于其高柔性，使其成为智能制造领域的一个重要研究方向。然而，机器人成形过程中局部变形机制复杂、支撑不足等问题往往导致成形件几何精度低，这是阻碍机器人成形在高精度制造中广泛应用的最关键因素之一。在这种情况下，本文提出了一个全面的方法，以提高几何精度的机器人成形。首先，针对机器人成形过程中几何精度这一核心问题，系统分析了机器人成形过程中几何误差的种类和产生原因。随后，将典型的精密控制策略及其相关研究分为工艺创新、工艺参数优化和刀具轨迹规划与补偿三个方面。还总结了与每种战略相关的当前挑战。最后，讨论了未来的研究方向，包括多机器人成形、人工智能、多源数据融合和数字孪生等先进技术。

{"title":"Improve geometric accuracy in robotic forming","authors":"Yanrong Zhang , Zeran Hou , Fan Yang , Bernd Kuhlenkötter , Antonio Sánchez Egea , Junying Min","doi":"10.1016/j.rcim.2026.103236","DOIUrl":"10.1016/j.rcim.2026.103236","url":null,"abstract":"<div><div>Robotic forming is a die-less manufacturing process that utilizes single or multiple industrial robots to incrementally deform metal sheets into complex, customized parts. Owing to its high flexibility, it has become an important research direction in the field of intelligent manufacturing due to its high flexibility. However, issues such as complex local deformation mechanisms and insufficient support during robotic forming often lead to low geometric accuracy in fabricated parts, which is one of the most critical factors hindering the broader application of robotic forming in high-precision manufacturing. In this context, this paper presents a comprehensive review of the methods for improving geometric accuracy in robotic forming. Firstly, to address the core issue of geometric accuracy, the categories and origins of geometric errors in robotic forming are systematically examined. Subsequently, typical precision control strategies and related research are categorized into three aspects: process innovation, optimization of process parameters, and tool path planning and compensation. Current challenges associated with each strategy are also summarized. Finally, potential future research directions are discussed, incorporating advanced technologies such as multi-robot forming, artificial intelligence, multi-source data fusion, and digital twins.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"100 ","pages":"Article 103236"},"PeriodicalIF":11.4,"publicationDate":"2026-01-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145956531","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 digital twin-driven deviation analysis approach for aircraft hydraulic pipelines assembly 飞机液压管路装配双驱动偏差数字分析方法

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

Robotics and Computer-integrated Manufacturing

Pub Date : 2026-01-10 DOI: 10.1016/j.rcim.2026.103234

Liang Chen , Yue Zhao , Aihua Su , Wenqiang Yang , Yu Zheng

The assembly deviation of aircraft hydraulic pipelines plays a critical role in ensuring flight stability. Nonetheless, the presence of multiple dynamic factors originating from various sources, which influence the final quality throughout the assembly process, poses significant challenges to conventional deviation analysis methods that rely on theoretical deviation calculation models with predetermined assembly parameters. To address these limitations, a digital twin (DT)-driven framework is proposed for the realtime assembly deviation analysis of aircraft hydraulic pipelines. First, a DT model encompassing the total factor information of the assembly process is constructed, comprising the resource, process and quality models. Then, a DT-driven assembly quality analysis method is presented. By integrating multidimensional data within a virtual environment and employing DT models to estimate assembly deviations in realtime, the quality of the entire pipeline system can be effectively analyzed. Finally, the efficacy of the proposed method is demonstrated through a case study involving the aircraft hydraulic pipeline assembly process.

飞机液压管路的装配偏差对飞机的飞行稳定性起着至关重要的作用。然而，在装配过程中，各种来源的多种动态因素会影响最终质量，这对传统的偏差分析方法提出了重大挑战，这些方法依赖于具有预定装配参数的理论偏差计算模型。针对这些局限性，提出了一种用于飞机液压管路装配偏差实时分析的数字孪生驱动框架。首先，构建了包含装配过程全要素信息的DT模型，包括资源模型、过程模型和质量模型。然后，提出了一种基于dt驱动的装配质量分析方法。通过集成虚拟环境中的多维数据，利用DT模型实时估计装配偏差，可以有效地分析整个管道系统的质量。最后，以飞机液压管路装配过程为例，验证了该方法的有效性。

{"title":"A digital twin-driven deviation analysis approach for aircraft hydraulic pipelines assembly","authors":"Liang Chen , Yue Zhao , Aihua Su , Wenqiang Yang , Yu Zheng","doi":"10.1016/j.rcim.2026.103234","DOIUrl":"10.1016/j.rcim.2026.103234","url":null,"abstract":"<div><div>The assembly deviation of aircraft hydraulic pipelines plays a critical role in ensuring flight stability. Nonetheless, the presence of multiple dynamic factors originating from various sources, which influence the final quality throughout the assembly process, poses significant challenges to conventional deviation analysis methods that rely on theoretical deviation calculation models with predetermined assembly parameters. To address these limitations, a digital twin (DT)-driven framework is proposed for the realtime assembly deviation analysis of aircraft hydraulic pipelines. First, a DT model encompassing the total factor information of the assembly process is constructed, comprising the resource, process and quality models. Then, a DT-driven assembly quality analysis method is presented. By integrating multidimensional data within a virtual environment and employing DT models to estimate assembly deviations in realtime, the quality of the entire pipeline system can be effectively analyzed. Finally, the efficacy of the proposed method is demonstrated through a case study involving the aircraft hydraulic pipeline assembly process.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"100 ","pages":"Article 103234"},"PeriodicalIF":11.4,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145940155","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 tool wear prediction model integrating mechanism-data fusion with transfer learning for machine tool digital twins 基于机制-数据融合和迁移学习的机床数字孪生刀具磨损预测模型

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

Robotics and Computer-integrated Manufacturing

Pub Date : 2026-01-09 DOI: 10.1016/j.rcim.2026.103235

Zenghui Wang , Guanghui Zhou , Chao Zhang , Wenhao Wang , Fengtian Chang , Yongrui Yu , Kun Li , Dan Zhao

Digital Twin enables high-fidelity data integration, dynamic state prediction, and adaptive decision-making in machining process. A robust tool wear monitoring (TWM) model is pivotal for ensuring data fidelity in digital twin systems and providing accurate evaluation of tool wear states. However, existing methods generally rely on idealized assumptions or high-quality datasets, while neglecting the physical wear mechanisms, exhibiting weak physical consistency and limited generalization under variable machining conditions. To overcome these challenges, this study proposes a TWM model that integrates the physics-informed Gaussian process regression with transfer learning within digital twin intelligent monitoring of milling process. Concretely, multi-domain features are extracted from monitoring signals and a high-correlation feature subspace is constructed by comprehensive correlation analysis. Then, a physics-based model is developed and embedded into the Gaussian process regression framework to provide prior knowledge of tool degradation. Furthermore, a two-stage transfer learning strategy is incorporated to facilitate effective cross-condition knowledge adaptation, significantly enhances the model’s generalization capability. Finally, the proposed method is validated on the PHM2010 dataset and milling experiments. Experimental results indicate that the method achieves RMSE between 0.20

μ m

and 0.63

μ m

and MAE between 0.14

μ m

and 0.39

μ m

across varying machining conditions, confirming its prediction accuracy and robust cross-condition adaptability. The proposed model provides reliable tool wear data for digital twin monitoring systems, facilitating intelligent decision-making and maintenance.

数字孪生实现了加工过程的高保真数据集成、动态状态预测和自适应决策。一个强大的刀具磨损监测（TWM）模型对于确保数字孪生系统中的数据保真度和提供刀具磨损状态的准确评估至关重要。然而，现有的方法通常依赖于理想化的假设或高质量的数据集，而忽略了物理磨损机理，在可变加工条件下表现出较弱的物理一致性和有限的泛化。为了克服这些挑战，本研究提出了一种TWM模型，该模型在铣削过程的数字孪生智能监测中集成了物理信息高斯过程回归和迁移学习。具体而言，从监测信号中提取多域特征，并通过综合相关分析构建高相关特征子空间。然后，开发了基于物理的模型并将其嵌入到高斯过程回归框架中，以提供工具退化的先验知识。此外，引入了两阶段迁移学习策略，实现了有效的跨条件知识自适应，显著提高了模型的泛化能力。最后，在PHM2010数据集和铣削实验上对该方法进行了验证。实验结果表明，该方法在不同加工条件下的RMSE在0.20 μm ~ 0.63 μm之间，MAE在0.14 μm ~ 0.39 μm之间，验证了该方法的预测精度和鲁棒的跨条件适应性。该模型为数字孪生监测系统提供了可靠的刀具磨损数据，便于智能决策和维护。

{"title":"A tool wear prediction model integrating mechanism-data fusion with transfer learning for machine tool digital twins","authors":"Zenghui Wang , Guanghui Zhou , Chao Zhang , Wenhao Wang , Fengtian Chang , Yongrui Yu , Kun Li , Dan Zhao","doi":"10.1016/j.rcim.2026.103235","DOIUrl":"10.1016/j.rcim.2026.103235","url":null,"abstract":"<div><div>Digital Twin enables high-fidelity data integration, dynamic state prediction, and adaptive decision-making in machining process. A robust tool wear monitoring (TWM) model is pivotal for ensuring data fidelity in digital twin systems and providing accurate evaluation of tool wear states. However, existing methods generally rely on idealized assumptions or high-quality datasets, while neglecting the physical wear mechanisms, exhibiting weak physical consistency and limited generalization under variable machining conditions. To overcome these challenges, this study proposes a TWM model that integrates the physics-informed Gaussian process regression with transfer learning within digital twin intelligent monitoring of milling process. Concretely, multi-domain features are extracted from monitoring signals and a high-correlation feature subspace is constructed by comprehensive correlation analysis. Then, a physics-based model is developed and embedded into the Gaussian process regression framework to provide prior knowledge of tool degradation. Furthermore, a two-stage transfer learning strategy is incorporated to facilitate effective cross-condition knowledge adaptation, significantly enhances the model’s generalization capability. Finally, the proposed method is validated on the PHM2010 dataset and milling experiments. Experimental results indicate that the method achieves RMSE between 0.20 <span><math><mrow><mi>μ</mi><mi>m</mi></mrow></math></span> and 0.63 <span><math><mrow><mi>μ</mi><mi>m</mi></mrow></math></span> and MAE between 0.14 <span><math><mrow><mi>μ</mi><mi>m</mi></mrow></math></span> and 0.39 <span><math><mrow><mi>μ</mi><mi>m</mi></mrow></math></span> across varying machining conditions, confirming its prediction accuracy and robust cross-condition adaptability. The proposed model provides reliable tool wear data for digital twin monitoring systems, facilitating intelligent decision-making and maintenance.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"100 ","pages":"Article 103235"},"PeriodicalIF":11.4,"publicationDate":"2026-01-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145940157","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