Robotics and Computer-integrated Manufacturing最新文献_第10页

Human-in-the-loop Multi-objective Bayesian Optimization for Directed Energy Deposition with in-situ monitoring 利用原位监测对定向能量沉积进行人在环多目标贝叶斯优化

IF 9.1 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Robotics and Computer-integrated Manufacturing

Pub Date : 2024-11-07 DOI: 10.1016/j.rcim.2024.102892

João Sousa , Armando Sousa , Frank Brueckner , Luís Paulo Reis , Ana Reis

Directed Energy Deposition (DED) is a free-form metal additive manufacturing process characterized as toolless, flexible, and energy-efficient compared to traditional processes. However, it is a complex system with a highly dynamic nature that presents challenges for modeling and optimization due to its multiphysics and multiscale characteristics. Additionally, multiple factors such as different machine setups and materials require extensive testing through single-track depositions, which can be time and resource-intensive. Single-track experiments are the foundation for establishing optimal initial parameters and comprehensively characterizing bead geometry, ensuring the accuracy and efficiency of computer-aided design and process quality validation. We digitized a DED setup using the Robot Operating System (ROS 2) and employed a thermal camera for real-time monitoring and evaluation to streamline the experimentation process. With the laser power and velocity as inputs, we optimized the dimensions and stability of the melt pool and evaluated different objective functions and approaches using a Response Surface Model (RSM). The three-objective approach achieved better rewards in all iterations and, when implemented in a real setup, allowed to reduce the number of experiments and shorten setup time. Our approach can minimize waste, increase the quality and reliability of DED, and enhance and simplify human-process interaction by leveraging the collaboration between human knowledge and model predictions.

定向能量沉积（DED）是一种自由形态的金属添加制造工艺，与传统工艺相比，它具有无工具、灵活和节能的特点。然而，由于其多物理和多尺度特性，它是一个具有高度动态性质的复杂系统，给建模和优化带来了挑战。此外，不同的机器设置和材料等多种因素要求通过单轨沉积进行大量测试，这可能会耗费大量时间和资源。单轨实验是建立最佳初始参数和全面表征微珠几何形状的基础，可确保计算机辅助设计和工艺质量验证的准确性和效率。我们使用机器人操作系统（ROS 2）将 DED 设置数字化，并使用热像仪进行实时监控和评估，以简化实验过程。以激光功率和速度作为输入，我们优化了熔池的尺寸和稳定性，并使用响应面模型（RSM）评估了不同的目标函数和方法。三目标方法在所有迭代中都获得了更好的回报，在实际设置中实施时，可以减少实验次数，缩短设置时间。我们的方法可以最大限度地减少浪费，提高 DED 的质量和可靠性，并通过利用人类知识和模型预测之间的协作来增强和简化人机交互。

{"title":"Human-in-the-loop Multi-objective Bayesian Optimization for Directed Energy Deposition with in-situ monitoring","authors":"João Sousa , Armando Sousa , Frank Brueckner , Luís Paulo Reis , Ana Reis","doi":"10.1016/j.rcim.2024.102892","DOIUrl":"10.1016/j.rcim.2024.102892","url":null,"abstract":"<div><div>Directed Energy Deposition (DED) is a free-form metal additive manufacturing process characterized as toolless, flexible, and energy-efficient compared to traditional processes. However, it is a complex system with a highly dynamic nature that presents challenges for modeling and optimization due to its multiphysics and multiscale characteristics. Additionally, multiple factors such as different machine setups and materials require extensive testing through single-track depositions, which can be time and resource-intensive. Single-track experiments are the foundation for establishing optimal initial parameters and comprehensively characterizing bead geometry, ensuring the accuracy and efficiency of computer-aided design and process quality validation. We digitized a DED setup using the Robot Operating System (ROS 2) and employed a thermal camera for real-time monitoring and evaluation to streamline the experimentation process. With the laser power and velocity as inputs, we optimized the dimensions and stability of the melt pool and evaluated different objective functions and approaches using a Response Surface Model (RSM). The three-objective approach achieved better rewards in all iterations and, when implemented in a real setup, allowed to reduce the number of experiments and shorten setup time. Our approach can minimize waste, increase the quality and reliability of DED, and enhance and simplify human-process interaction by leveraging the collaboration between human knowledge and model predictions.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"92 ","pages":"Article 102892"},"PeriodicalIF":9.1,"publicationDate":"2024-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142637685","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A two-stage dynamic model updating method for the FRF prediction of the robotized mobile machining equipment 机器人移动加工设备 FRF 预测的两阶段动态模型更新方法

IF 9.1 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Robotics and Computer-integrated Manufacturing

Pub Date : 2024-11-06 DOI: 10.1016/j.rcim.2024.102895

Zijian Ma , Fugui Xie , Xin-Jun Liu

In the coming decades, robotized mobile machining equipment (RMME) is possible to evolve as a new branch of machine tools due to its exceptional flexibility. The frequency response function (FRF) serves as a theoretical foundation in controlling the vibration deformations that significantly limit the material removal efficiency of RMME. Model updating, aimed at minimizing errors between the theoretical model and the physical prototype, is essential to predict the FRF accurately. However, updating the dynamic model of RMME, characterized by non-mechanical boundary conditions, complex lightweight components, and low-stiffness structures, presents difficulties in computational efficiency and updating posture-dependent parameters. To solve these issues, the prediction error generation mechanism is first analyzed to confirm the error types that need to be eliminated in model updating. A two-stage model updating method that can separately update the robot structure and boundary-related parameters is proposed to rapidly update the dynamic model under various machining tasks. The interface reduction technique that can decrease the model order is introduced to reduce the computational consumption, and an approach to fast update such interface reduction substructures is put forward to avoid the reiterative model reduction during updating. An updating method for posture-dependent parameters based on multi-objective optimization is designed to control the multiple solution issues by generating many feasible solutions, ensuring the prediction effects for non-updated postures. The experimental results indicate that updating structural parameters in stage I and adsorption surface parameters in stage II results in mean error reduction percentages of 58.79 % and 48.30 %, respectively. Additionally, the natural frequencies and the mode shapes can also be predicted by the updated model. Comparative analysis with various controlled groups confirms the advantages of utilizing posture-dependent parameters in prediction accuracy and adopting the two-stage model updating method in efficiency. The proposed method can also be applied to other RMMEs.

在未来几十年中，机器人移动加工设备（RMME）因其卓越的灵活性有可能发展成为机床的一个新分支。频率响应函数（FRF）是控制振动变形的理论基础，而振动变形极大地限制了 RMME 的材料去除效率。模型更新旨在最大限度地减少理论模型与物理原型之间的误差，对于准确预测频率响应函数至关重要。然而，RMME 的动态模型具有非机械边界条件、复杂的轻质部件和低刚度结构等特点，更新模型在计算效率和更新与姿态相关的参数方面存在困难。为解决这些问题，首先分析了预测误差的产生机制，以确认模型更新中需要消除的误差类型。提出了一种可分别更新机器人结构和边界相关参数的两阶段模型更新方法，以快速更新各种加工任务下的动态模型。引入了可减少模型阶数的界面缩减技术以降低计算消耗，并提出了快速更新这种界面缩减子结构的方法，以避免更新过程中的重复模型缩减。设计了一种基于多目标优化的姿态相关参数更新方法，通过生成多个可行解来控制多解问题，确保未更新姿态的预测效果。实验结果表明，在第一阶段更新结构参数和在第二阶段更新吸附面参数后，平均误差降低率分别为 58.79 % 和 48.30 %。此外，更新后的模型还能预测固有频率和模态振型。与不同对照组的对比分析证实了利用姿态相关参数在预测精度上的优势，以及采用两阶段模型更新方法在效率上的优势。所提出的方法也可应用于其他 RMME。

{"title":"A two-stage dynamic model updating method for the FRF prediction of the robotized mobile machining equipment","authors":"Zijian Ma , Fugui Xie , Xin-Jun Liu","doi":"10.1016/j.rcim.2024.102895","DOIUrl":"10.1016/j.rcim.2024.102895","url":null,"abstract":"<div><div>In the coming decades, robotized mobile machining equipment (RMME) is possible to evolve as a new branch of machine tools due to its exceptional flexibility. The frequency response function (FRF) serves as a theoretical foundation in controlling the vibration deformations that significantly limit the material removal efficiency of RMME. Model updating, aimed at minimizing errors between the theoretical model and the physical prototype, is essential to predict the FRF accurately. However, updating the dynamic model of RMME, characterized by non-mechanical boundary conditions, complex lightweight components, and low-stiffness structures, presents difficulties in computational efficiency and updating posture-dependent parameters. To solve these issues, the prediction error generation mechanism is first analyzed to confirm the error types that need to be eliminated in model updating. A two-stage model updating method that can separately update the robot structure and boundary-related parameters is proposed to rapidly update the dynamic model under various machining tasks. The interface reduction technique that can decrease the model order is introduced to reduce the computational consumption, and an approach to fast update such interface reduction substructures is put forward to avoid the reiterative model reduction during updating. An updating method for posture-dependent parameters based on multi-objective optimization is designed to control the multiple solution issues by generating many feasible solutions, ensuring the prediction effects for non-updated postures. The experimental results indicate that updating structural parameters in stage I and adsorption surface parameters in stage II results in mean error reduction percentages of 58.79 % and 48.30 %, respectively. Additionally, the natural frequencies and the mode shapes can also be predicted by the updated model. Comparative analysis with various controlled groups confirms the advantages of utilizing posture-dependent parameters in prediction accuracy and adopting the two-stage model updating method in efficiency. The proposed method can also be applied to other RMMEs.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"92 ","pages":"Article 102895"},"PeriodicalIF":9.1,"publicationDate":"2024-11-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142593790","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

Variable impedance control on contact-rich manipulation of a collaborative industrial mobile manipulator: An imitation learning approach 可变阻抗控制协作式工业移动机械手的丰富接触操作：模仿学习法

IF 9.1 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Robotics and Computer-integrated Manufacturing

Pub Date : 2024-11-05 DOI: 10.1016/j.rcim.2024.102896

Zhengxue Zhou, Xingyu Yang, Xuping Zhang

Variable impedance control (VIC) endows robots with the ability to adjust their compliance, enhancing safety and adaptability in contact-rich tasks. However, determining suitable variable impedance parameters for specific tasks remains challenging. To address this challenge, this paper proposes an imitation learning-based VIC policy that employs observations integrated with RGBD and force/torque (F/T) data enabling a collaborative mobile manipulator to execute contact-rich tasks by learning from human demonstrations. The VIC policy is learned through training the robot in a customized simulation environment, utilizing an inverse reinforcement learning (IRL) algorithm. High-dimensional demonstration data is encoded by integrating a 16-layer convolutional neural network (CNN) into the IRL environment. To minimize the sim-to-real gap, contact dynamic parameters in the training environment are calibrated. Then, the learning-based VIC policy is comprehensively trained in the customized environment and its transferability is validated through an industrial production case involving a high precision peg-in-hole task using a collaborative mobile manipulator. The training and testing results indicate that the proposed imitation learning-based VIC policy ensures robust performance for contact-rich tasks.

可变阻抗控制（VIC）赋予机器人调整顺应性的能力，提高了机器人在多接触任务中的安全性和适应性。然而，为特定任务确定合适的可变阻抗参数仍然具有挑战性。为了应对这一挑战，本文提出了一种基于模仿学习的可变阻抗控制策略，该策略采用与 RGBD 和力/力矩（F/T）数据相结合的观察结果，使协作式移动机械手能够通过学习人类示范来执行接触丰富的任务。VIC 策略是通过在定制的模拟环境中训练机器人，并利用反强化学习（IRL）算法来学习的。通过将 16 层卷积神经网络 (CNN) 集成到 IRL 环境中，对高维演示数据进行编码。为了尽量缩小模拟与真实之间的差距，对训练环境中的接触动态参数进行了校准。然后，在定制环境中全面训练基于学习的 VIC 策略，并通过使用协作移动机械手执行高精度钉入孔任务的工业生产案例验证其可移植性。训练和测试结果表明，所提出的基于模仿学习的虚拟集成电路策略可确保在接触丰富的任务中表现稳健。

{"title":"Variable impedance control on contact-rich manipulation of a collaborative industrial mobile manipulator: An imitation learning approach","authors":"Zhengxue Zhou, Xingyu Yang, Xuping Zhang","doi":"10.1016/j.rcim.2024.102896","DOIUrl":"10.1016/j.rcim.2024.102896","url":null,"abstract":"<div><div>Variable impedance control (VIC) endows robots with the ability to adjust their compliance, enhancing safety and adaptability in contact-rich tasks. However, determining suitable variable impedance parameters for specific tasks remains challenging. To address this challenge, this paper proposes an imitation learning-based VIC policy that employs observations integrated with RGBD and force/torque (F/T) data enabling a collaborative mobile manipulator to execute contact-rich tasks by learning from human demonstrations. The VIC policy is learned through training the robot in a customized simulation environment, utilizing an inverse reinforcement learning (IRL) algorithm. High-dimensional demonstration data is encoded by integrating a 16-layer convolutional neural network (CNN) into the IRL environment. To minimize the sim-to-real gap, contact dynamic parameters in the training environment are calibrated. Then, the learning-based VIC policy is comprehensively trained in the customized environment and its transferability is validated through an industrial production case involving a high precision peg-in-hole task using a collaborative mobile manipulator. The training and testing results indicate that the proposed imitation learning-based VIC policy ensures robust performance for contact-rich tasks.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"92 ","pages":"Article 102896"},"PeriodicalIF":9.1,"publicationDate":"2024-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142585486","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Optimal model-based path planning for the robotic manipulation of deformable linear objects 基于模型的机器人操纵可变形线性物体的最优路径规划

IF 9.1 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Robotics and Computer-integrated Manufacturing

Pub Date : 2024-11-04 DOI: 10.1016/j.rcim.2024.102891

Andrea Monguzzi, Tommaso Dotti, Lorenzo Fattorelli, Andrea Maria Zanchettin, Paolo Rocco

The robotic manipulation of deformable linear objects (DLOs), such as cables, is a valuable yet complex skill. In particular, to realize tasks like cable routing and wire harness assembly, it is required that two robotic arms, grasping the ends of a DLO, move it from an initial shape to a final one where cable assembly can be performed. The manipulation must be performed following a collision-free path and avoiding stretching and excessively deforming it. We address this problem by proposing an optimal model-based path planning strategy. Specifically, a hierarchical optimization strategy is defined to perform path planning, exploiting a mass–spring DLO dynamic model that we enhance to handle a generic equilibrium condition for the DLO. Furthermore, we model the interaction of the DLO with objects like clips used in assembly operations. We also deal with the estimation of the DLO stiffness to properly tune the model parameters. The effectiveness of our methodology is assessed via experimental tests, where a dual-arm robot executes the planned paths manipulating several DLOs with different mechanical properties. Finally, the method is exploited to execute a wire harness assembly task.

机器人操纵电缆等可变形线性物体（DLO）是一项宝贵而复杂的技能。特别是，要完成电缆布线和线束组装等任务，需要两个机械臂抓住 DLO 的两端，将其从初始形状移动到最终形状，以便进行电缆组装。操作必须按照无碰撞路径进行，并避免拉伸和过度变形。针对这一问题，我们提出了一种基于模型的最优路径规划策略。具体来说，我们定义了一种分层优化策略来执行路径规划，该策略利用了质量-弹簧 DLO 动态模型，我们对该模型进行了改进，以处理 DLO 的通用平衡条件。此外，我们还模拟了 DLO 与装配操作中使用的夹子等物体之间的相互作用。我们还对 DLO 的刚度进行了估计，以适当调整模型参数。我们通过实验测试评估了我们方法的有效性，在实验测试中，双臂机器人通过操纵具有不同机械性能的多个 DLO 执行计划路径。最后，我们利用该方法执行了一项线束装配任务。

{"title":"Optimal model-based path planning for the robotic manipulation of deformable linear objects","authors":"Andrea Monguzzi, Tommaso Dotti, Lorenzo Fattorelli, Andrea Maria Zanchettin, Paolo Rocco","doi":"10.1016/j.rcim.2024.102891","DOIUrl":"10.1016/j.rcim.2024.102891","url":null,"abstract":"<div><div>The robotic manipulation of deformable linear objects (DLOs), such as cables, is a valuable yet complex skill. In particular, to realize tasks like cable routing and wire harness assembly, it is required that two robotic arms, grasping the ends of a DLO, move it from an initial shape to a final one where cable assembly can be performed. The manipulation must be performed following a collision-free path and avoiding stretching and excessively deforming it. We address this problem by proposing an optimal model-based path planning strategy. Specifically, a hierarchical optimization strategy is defined to perform path planning, exploiting a mass–spring DLO dynamic model that we enhance to handle a generic equilibrium condition for the DLO. Furthermore, we model the interaction of the DLO with objects like clips used in assembly operations. We also deal with the estimation of the DLO stiffness to properly tune the model parameters. The effectiveness of our methodology is assessed via experimental tests, where a dual-arm robot executes the planned paths manipulating several DLOs with different mechanical properties. Finally, the method is exploited to execute a wire harness assembly task.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"92 ","pages":"Article 102891"},"PeriodicalIF":9.1,"publicationDate":"2024-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142577804","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

Design synthesis, modeling, control strategies, and fabrication methods of compliant grippers for micromanipulation and micromanipulator: A comprehensive review 用于微操作和微机械手的顺应式机械手的设计合成、建模、控制策略和制造方法：全面综述

IF 9.1 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Robotics and Computer-integrated Manufacturing

Pub Date : 2024-11-02 DOI: 10.1016/j.rcim.2024.102893

Hieu Giang Le , Nhat Linh Ho , Thanh-Phong Dao

In robotic and automation industry, micromanipulation and micromanipulator have recognized significant advancements due to they are involved in handling of micro-sized parts from a few to hundreds of micrometers. In order to perform such precise grasping tasks, compliant grippers have been increasingly developed, and they have critically significant contributions in the high precision micromanipulation and micromanipulator. This article aims to present a comprehensive review on the state-of-the-art development of compliant grippers. This review focuses on design synthesis, modeling methods, control strategies, and fabrication technologies for compliant grippers. Each section is deeply analyzed and discussed. This paper identifies ongoing challenges and outlines future prospects for developing compliant grippers. The achieved results of this review can provide and inspire helpful insights in ultra-high precision micromanipulation and micromanipulator.

在机器人和自动化行业中，微机械和微机械手因其涉及处理从几微米到几百微米的微小零件而取得了长足的进步。为了完成如此精确的抓取任务，顺应式机械手得到了越来越多的开发，它们在高精度微机械和微机械手领域做出了重要贡献。本文旨在对顺应式机械手的最新发展进行全面综述。综述的重点是顺应式机械手的设计合成、建模方法、控制策略和制造技术。每个部分都进行了深入分析和讨论。本文指出了当前面临的挑战，并概述了顺应式机械手的未来发展前景。本综述所取得的成果可为超高精度微机械和微机械手提供有益的启发。

引用次数: 0

Self-training-based approach with improved XGBoost for aluminum alloy casting quality prediction 基于自训练和改进 XGBoost 的铝合金铸件质量预测方法

IF 9.1 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Robotics and Computer-integrated Manufacturing

Pub Date : 2024-11-02 DOI: 10.1016/j.rcim.2024.102890

Haonan Wang , Quanzhi Sun , Jun Wu , Xuxia Zhang , Weipeng Liu , Tao Peng , Renzhong Tang

The revolutionary advances in integrated components in current automotive industry have led to a sharply rising demand for aluminum alloy castings. Targeted quality inspection is thus proposed for components manufacturers to achieve high responsiveness and low operational cost. This suggests casting machine manufacturers to integrate advanced quality prediction functions into the next generation of intelligent casting machines. However, acquiring ample quality inspection data is essential for implementing such functions, which is often challenging, if not infeasible, due to practical issues such as data proprietorship or privacy. Self-training is a good candidate for dealing with scarce labeled data, and XGBoost is commonly used as the base classifier. However, misclassification of unlabeled data happens using XGBoost, which could lead to incorrect pseudo-label assignments, eventually resulting in higher misclassification rate. To address this challenge, a self-training and improved XGBoost-based aluminum alloy casting quality prediction approach is proposed. This approach integrates the classification loss of unlabeled data in the objective function as a new regularization term and considers first and second partial derivatives of the classification loss function for unlabeled data in the leaf node's weight score. The proposed approach penalizes those classification models that misclassify unlabeled data, thereby improves quality prediction performance. To evaluate the effectiveness of our approach, a casting machine manufacturer was collaborated to conduct a case study. The results on three-type casting quality prediction demonstrate that our approach could achieve an accuracy, precision, recall and F1 score of 93.2 %, 90 %, 64.2 %, and 0.75, respectively, outperforming all compared approaches. The approach supports casting machine manufacturers to pre-train a casting quality prediction models with scarce labeled data, enabling swift deployment and customization for targeted quality inspection.

当前，汽车工业中集成组件的革命性进步导致对铝合金铸件的需求急剧上升。因此，为了实现高响应速度和低运营成本，零部件制造商需要进行有针对性的质量检测。这建议铸造机制造商在下一代智能铸造机中集成先进的质量预测功能。然而，要实现这些功能，获取充足的质量检测数据是必不可少的，但由于数据所有权或隐私等实际问题，这往往具有挑战性，甚至是不可行的。自我训练是处理稀缺标记数据的好方法，XGBoost 通常被用作基本分类器。然而，使用 XGBoost 时会出现对未标记数据的误分类，这可能会导致错误的伪标签分配，最终导致更高的误分类率。为了应对这一挑战，我们提出了一种基于 XGBoost 的自训练改进型铝合金铸件质量预测方法。该方法将目标函数中未标注数据的分类损失作为一个新的正则项，并在叶节点的权重得分中考虑未标注数据的分类损失函数的一阶和二阶偏导数。所提出的方法可以惩罚那些对未标注数据进行错误分类的分类模型，从而提高质量预测性能。为了评估我们方法的有效性，我们与一家铸造机制造商合作进行了案例研究。三类铸件质量预测结果表明，我们的方法在准确度、精确度、召回率和 F1 分数上分别达到了 93.2%、90%、64.2% 和 0.75，优于所有比较方法。该方法支持铸造机制造商利用稀缺的标注数据预先训练铸造质量预测模型，从而实现快速部署和定制目标质量检测。

{"title":"Self-training-based approach with improved XGBoost for aluminum alloy casting quality prediction","authors":"Haonan Wang , Quanzhi Sun , Jun Wu , Xuxia Zhang , Weipeng Liu , Tao Peng , Renzhong Tang","doi":"10.1016/j.rcim.2024.102890","DOIUrl":"10.1016/j.rcim.2024.102890","url":null,"abstract":"<div><div>The revolutionary advances in integrated components in current automotive industry have led to a sharply rising demand for aluminum alloy castings. Targeted quality inspection is thus proposed for components manufacturers to achieve high responsiveness and low operational cost. This suggests casting machine manufacturers to integrate advanced quality prediction functions into the next generation of intelligent casting machines. However, acquiring ample quality inspection data is essential for implementing such functions, which is often challenging, if not infeasible, due to practical issues such as data proprietorship or privacy. Self-training is a good candidate for dealing with scarce labeled data, and XGBoost is commonly used as the base classifier. However, misclassification of unlabeled data happens using XGBoost, which could lead to incorrect pseudo-label assignments, eventually resulting in higher misclassification rate. To address this challenge, a self-training and improved XGBoost-based aluminum alloy casting quality prediction approach is proposed. This approach integrates the classification loss of unlabeled data in the objective function as a new regularization term and considers first and second partial derivatives of the classification loss function for unlabeled data in the leaf node's weight score. The proposed approach penalizes those classification models that misclassify unlabeled data, thereby improves quality prediction performance. To evaluate the effectiveness of our approach, a casting machine manufacturer was collaborated to conduct a case study. The results on three-type casting quality prediction demonstrate that our approach could achieve an accuracy, precision, recall and F1 score of 93.2 %, 90 %, 64.2 %, and 0.75, respectively, outperforming all compared approaches. The approach supports casting machine manufacturers to pre-train a casting quality prediction models with scarce labeled data, enabling swift deployment and customization for targeted quality inspection.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"92 ","pages":"Article 102890"},"PeriodicalIF":9.1,"publicationDate":"2024-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142571631","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

Adaptively sampled distance functions: A unifying digital twin representation for advanced manufacturing 自适应采样距离函数：先进制造业的统一数字孪生表示法

IF 9.1 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Robotics and Computer-integrated Manufacturing

Pub Date : 2024-10-28 DOI: 10.1016/j.rcim.2024.102877

Sam Pratt , Tadeusz Kosmal , Christopher Williams

Digital twin tools for additive manufacturing (AM) are constrained by the underlying representations of component geometry that are currently in wide use. Mesh, voxel, and parametric surface representations require numerous conversions to intermediate representations at multiple points throughout the processing chain. Each conversion introduces additional error in the geometric representation and complicates comparison of in-situ process sensor data to the as-designed component. Additionally, the limited interoperability of the various representations produced throughout the process chain limit the insights available from current digital twin tools. We introduce a novel framework based on a unifying geometric representation that serves the complete AM digital thread. The presented GPU-accelerated, adaptively sampled distance function (ASDF) framework serves as a foundation for component design and path planning tools, especially for real-time path planning in AM, as well as provides a baseline representation of geometry from control systems, and enables rapid comparison of in-situ sensor data to the as-designed model without intermediate conversion, greatly reducing the burden of reducing such data to usable process insights.

用于增材制造（AM）的数字孪生工具受到目前广泛使用的组件几何图形底层表示法的限制。网格、体素和参数化曲面表示法需要在整个加工链的多个环节进行大量的中间表示法转换。每次转换都会在几何表示法中引入额外的误差，并使原位工艺传感器数据与设计组件的比较变得复杂。此外，在整个加工链中产生的各种表征的互操作性有限，限制了当前数字孪生工具的洞察力。我们引入了一个基于统一几何表示法的新型框架，该表示法可用于整个 AM 数字线程。所介绍的 GPU 加速自适应采样距离函数（ASDF）框架可作为组件设计和路径规划工具的基础，尤其适用于 AM 中的实时路径规划，还可提供来自控制系统的几何基准表示法，并可将现场传感器数据与设计模型进行快速比较，而无需进行中间转换，从而大大减轻了将此类数据还原为可用工艺见解的负担。

{"title":"Adaptively sampled distance functions: A unifying digital twin representation for advanced manufacturing","authors":"Sam Pratt , Tadeusz Kosmal , Christopher Williams","doi":"10.1016/j.rcim.2024.102877","DOIUrl":"10.1016/j.rcim.2024.102877","url":null,"abstract":"<div><div>Digital twin tools for additive manufacturing (AM) are constrained by the underlying representations of component geometry that are currently in wide use. Mesh, voxel, and parametric surface representations require numerous conversions to intermediate representations at multiple points throughout the processing chain. Each conversion introduces additional error in the geometric representation and complicates comparison of <em>in-situ</em> process sensor data to the as-designed component. Additionally, the limited interoperability of the various representations produced throughout the process chain limit the insights available from current digital twin tools. We introduce a novel framework based on a unifying geometric representation that serves the complete AM digital thread. The presented GPU-accelerated, adaptively sampled distance function (ASDF) framework serves as a foundation for component design and path planning tools, especially for real-time path planning in AM, as well as provides a baseline representation of geometry from control systems, and enables rapid comparison of <em>in-situ</em> sensor data to the as-designed model without intermediate conversion, greatly reducing the burden of reducing such data to usable process insights.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"92 ","pages":"Article 102877"},"PeriodicalIF":9.1,"publicationDate":"2024-10-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142534212","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

Trajectory error compensation method for grinding robots based on kinematic calibration and joint variable prediction 基于运动学校准和关节变量预测的打磨机器人轨迹误差补偿方法

IF 9.1 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Robotics and Computer-integrated Manufacturing

Pub Date : 2024-10-25 DOI: 10.1016/j.rcim.2024.102889

Kaiwei Ma , Fengyu Xu , Qingyu Xu , Shuang Gao , Guo-Ping Jiang

Trajectory accuracy, a crucial metric in assessing the dynamic performance of grinding robots, is influenced by the uncertain movement of the tool center point, directly impacting the surface quality of processed workpieces. This article introduces an innovative method for compensating trajectory errors. Initially, a strategy for error compensation is derived using differential kinematics theory. Subsequently, a robot kinematic calibration method utilizing ring particle swarm optimization (RPSO) is proposed to address static errors in the grinding robot. Simultaneously, a method for predicting robot joint variables based on a dual-channel feedforward neural network (DCFNN) is designed to mitigate dynamic errors. Finally, a simulation platform is developed to validate the proposed method. Simulation analysis using extensive data demonstrates an 89.3% improvement in absolute position accuracy and a 74.2% reduction in error fluctuation range, outperforming sparrow search algorithm (SSA), improved mayfly algorithm (IMA), multi-representation integrated predictive neural network (MRIPNN), etc. Algorithmic comparison reveals that kinematic calibration significantly reduces the average trajectory error, while joint variable prediction notably minimizes error fluctuation. Validation through trajectory straightness testing and a 3D printing propeller grinding experiment achieves a trajectory straightness of 0.2425 mm. Implementing this method enables achieving 86.1% surface machining allowance within tolerance, making it an optimal solution for grinding robots.

轨迹精度是评估打磨机器人动态性能的关键指标，它受到刀具中心点不确定运动的影响，直接影响加工工件的表面质量。本文介绍了一种创新的轨迹误差补偿方法。首先，利用微分运动学理论推导出一种误差补偿策略。随后，提出了一种利用环形粒子群优化（RPSO）的机器人运动学校准方法，以解决打磨机器人的静态误差问题。同时，设计了一种基于双通道前馈神经网络（DCFNN）的机器人关节变量预测方法，以减少动态误差。最后，开发了一个仿真平台来验证所提出的方法。利用大量数据进行的仿真分析表明，绝对位置精度提高了 89.3%，误差波动范围缩小了 74.2%，优于麻雀搜索算法（SSA）、改进的蜉蝣算法（IMA）、多表征集成预测神经网络（MRIPNN）等。通过算法比较发现，运动校准能显著降低平均轨迹误差，而联合变量预测则能显著减少误差波动。通过轨迹直线度测试和 3D 打印螺旋桨研磨实验验证，轨迹直线度达到 0.2425 毫米。采用这种方法后，表面加工余量在公差范围内达到了 86.1%，成为打磨机器人的最佳解决方案。

{"title":"Trajectory error compensation method for grinding robots based on kinematic calibration and joint variable prediction","authors":"Kaiwei Ma , Fengyu Xu , Qingyu Xu , Shuang Gao , Guo-Ping Jiang","doi":"10.1016/j.rcim.2024.102889","DOIUrl":"10.1016/j.rcim.2024.102889","url":null,"abstract":"<div><div>Trajectory accuracy, a crucial metric in assessing the dynamic performance of grinding robots, is influenced by the uncertain movement of the tool center point, directly impacting the surface quality of processed workpieces. This article introduces an innovative method for compensating trajectory errors. Initially, a strategy for error compensation is derived using differential kinematics theory. Subsequently, a robot kinematic calibration method utilizing ring particle swarm optimization (RPSO) is proposed to address static errors in the grinding robot. Simultaneously, a method for predicting robot joint variables based on a dual-channel feedforward neural network (DCFNN) is designed to mitigate dynamic errors. Finally, a simulation platform is developed to validate the proposed method. Simulation analysis using extensive data demonstrates an 89.3% improvement in absolute position accuracy and a 74.2% reduction in error fluctuation range, outperforming sparrow search algorithm (SSA), improved mayfly algorithm (IMA), multi-representation integrated predictive neural network (MRIPNN), etc. Algorithmic comparison reveals that kinematic calibration significantly reduces the average trajectory error, while joint variable prediction notably minimizes error fluctuation. Validation through trajectory straightness testing and a 3D printing propeller grinding experiment achieves a trajectory straightness of 0.2425 mm. Implementing this method enables achieving 86.1% surface machining allowance within tolerance, making it an optimal solution for grinding robots.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"92 ","pages":"Article 102889"},"PeriodicalIF":9.1,"publicationDate":"2024-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142534211","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 hybrid model in a nonlinear disturbance observer for improving compliance error compensation of robotic machining 非线性扰动观测器中的混合模型，用于改进机器人加工的顺应性误差补偿

IF 9.1 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Robotics and Computer-integrated Manufacturing

Pub Date : 2024-10-24 DOI: 10.1016/j.rcim.2024.102887

Ali Khishtan , Seong Hyeon Kim , Jihyun Lee

The joint deflection of robots in machining degrades product accuracy. Compliance error compensation has been investigated to reduce the static deflection of robotic machining. The challenge in compliance error compensation is accurately measuring the deflection or cutting force. External sensors have been used to measure them in robotic machining, but it is not practical. The authors proposed a nonlinear disturbance observer to indirectly measure the cutting force online in robotic machining in the previous study. The observer, however, needs to utilize the robot model that includes characteristics of high nonlinearity, uncertainty, and high dynamic variation for different robot postures. After investigating these challenges of modeling, this paper proposes a hybrid modeling approach combining a physics-based model with a new empirical friction model, and a data-driven model to accurately estimate the cutting force while minimizing the error of the robot's mathematical model. The joint torque calculated from the hybrid model can cover the effect of joints' postures and speeds on the varying dynamic in its workspace. Real-time optimization just before cutting is also proposed to adapt to the real-time joint's motion conditions. The experimental results from aluminum multi-axis cutting show that the estimated cutting force via the nonlinear disturbance observer based on the proposed hybrid modeling approach can improve its accuracy up to 45% and 74% in the x and y directions respectively, compared to the physics-based modeling approach. The deflection of the tool center point can be compensated by using a compliance error compensation method up to 79.1% and 75.4% in the x and y directions, respectively, at 0.5 mm/s feed rate, and up to 77.2% and 78.9% at 3 mm/s feed rate. Consequently, the approaches developed in this paper can solve the problems of conventional robot modeling and improve the accuracy of robot machining.

机器人在加工过程中的关节挠度会降低产品精度。为了减少机器人加工中的静态挠度，人们对顺应误差补偿进行了研究。顺应性误差补偿的难点在于精确测量挠度或切削力。在机器人加工中，外部传感器被用来测量它们，但这并不实用。作者在之前的研究中提出了一种非线性干扰观测器，用于间接在线测量机器人加工中的切削力。然而，该观测器需要利用机器人模型，而机器人模型包括高非线性、不确定性和不同机器人姿态下的高动态变化等特点。在研究了建模所面临的这些挑战后，本文提出了一种混合建模方法，将基于物理的模型与新的经验摩擦模型和数据驱动模型相结合，在精确估算切削力的同时，最大限度地减小机器人数学模型的误差。混合模型计算出的关节扭矩可以涵盖关节姿态和速度对其工作空间内动态变化的影响。此外，还提出了切割前的实时优化，以适应关节的实时运动条件。铝材多轴切削的实验结果表明，与基于物理的建模方法相比，通过基于混合建模方法的非线性扰动观测器估算的切削力在 x 和 y 方向的精度分别提高了 45% 和 74%。在进给速度为 0.5 mm/s 的情况下，使用顺应性误差补偿方法，刀具中心点的偏移在 x 和 y 方向的补偿率分别可达 79.1% 和 75.4%；在进给速度为 3 mm/s 的情况下，补偿率分别可达 77.2% 和 78.9%。因此，本文开发的方法可以解决传统机器人建模的问题，提高机器人加工的精度。

{"title":"A hybrid model in a nonlinear disturbance observer for improving compliance error compensation of robotic machining","authors":"Ali Khishtan , Seong Hyeon Kim , Jihyun Lee","doi":"10.1016/j.rcim.2024.102887","DOIUrl":"10.1016/j.rcim.2024.102887","url":null,"abstract":"<div><div>The joint deflection of robots in machining degrades product accuracy. Compliance error compensation has been investigated to reduce the static deflection of robotic machining. The challenge in compliance error compensation is accurately measuring the deflection or cutting force. External sensors have been used to measure them in robotic machining, but it is not practical. The authors proposed a nonlinear disturbance observer to indirectly measure the cutting force online in robotic machining in the previous study. The observer, however, needs to utilize the robot model that includes characteristics of high nonlinearity, uncertainty, and high dynamic variation for different robot postures. After investigating these challenges of modeling, this paper proposes a hybrid modeling approach combining a physics-based model with a new empirical friction model, and a data-driven model to accurately estimate the cutting force while minimizing the error of the robot's mathematical model. The joint torque calculated from the hybrid model can cover the effect of joints' postures and speeds on the varying dynamic in its workspace. Real-time optimization just before cutting is also proposed to adapt to the real-time joint's motion conditions. The experimental results from aluminum multi-axis cutting show that the estimated cutting force via the nonlinear disturbance observer based on the proposed hybrid modeling approach can improve its accuracy up to 45% and 74% in the <em>x</em> and <em>y</em> directions respectively, compared to the physics-based modeling approach. The deflection of the tool center point can be compensated by using a compliance error compensation method up to 79.1% and 75.4% in the <em>x</em> and <em>y</em> directions, respectively, at 0.5 <em>mm/s</em> feed rate, and up to 77.2% and 78.9% at 3 <em>mm/s</em> feed rate. Consequently, the approaches developed in this paper can solve the problems of conventional robot modeling and improve the accuracy of robot machining.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"92 ","pages":"Article 102887"},"PeriodicalIF":9.1,"publicationDate":"2024-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142534213","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

Tool breakage monitoring driven by the real-time predicted spindle cutting torque using spindle servo signals 利用主轴伺服信号实时预测主轴切削扭矩，监测刀具破损情况

IF 9.1 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Robotics and Computer-integrated Manufacturing

Pub Date : 2024-10-15 DOI: 10.1016/j.rcim.2024.102888

Yinghao Cheng , Yingguang Li , Guangxu Li , Xu Liu , Jinyu Xia , Changqing Liu , Xiaozhong Hao

Monitoring tool breakage during computer numerical control machining is essential to ensure machining quality and equipment safety. In consideration of the low cost in long-term use and the non-invasiveness to workspace, using servo signals of machine tools to monitor tool breakage has been viewed as the solution that has great potential to be applied in real industry. However, because machine tool servo signals can only partially and indirectly reflect tool conditions, the accuracy and reliability of existing methods still need to be improved. To overcome this challenge, a novel two-step data-driven tool breakage monitoring method using spindle servo signals is proposed. Since spindle cutting torque is acknowledged as one of the most effective and reliable physical signals for detecting tool breakage, it is introduced as the key intermediate variable from spindle servo signals to tool conditions. The monitored spindle servo signals are used to predict the spindle cutting torque in real time based on a long short-term memory neural network, and then the predicted spindle cutting torque is used to detect tool breakage based on a one-dimensional convolutional neural network. The experimental results show that the proposed method can accurately predict the spindle cutting torque for normal tools and broken tools. Compared with the tool breakage monitoring methods that directly use spindle servo signals, the proposed method has higher detection accuracy and more reliable detection results, and the performance is more stable when increasing the detection frequency and decreasing training data.

监控计算机数控加工过程中的刀具破损对于确保加工质量和设备安全至关重要。考虑到长期使用的低成本和对工作空间的非侵入性，利用机床伺服信号监测刀具破损一直被视为在实际工业中具有巨大应用潜力的解决方案。然而，由于机床伺服信号只能部分和间接地反映刀具状况，现有方法的准确性和可靠性仍有待提高。为了克服这一难题，本文提出了一种利用主轴伺服信号的新型两步式数据驱动刀具破损监测方法。由于主轴切削扭矩被认为是检测刀具破损最有效、最可靠的物理信号之一，因此被引入作为从主轴伺服信号到刀具状况的关键中间变量。基于长短期记忆神经网络，利用监测到的主轴伺服信号实时预测主轴切削扭矩，然后基于一维卷积神经网络利用预测到的主轴切削扭矩检测刀具破损情况。实验结果表明，所提出的方法可以准确预测正常刀具和破损刀具的主轴切削扭矩。与直接使用主轴伺服信号的刀具破损监测方法相比，所提出的方法具有更高的检测精度和更可靠的检测结果，并且在增加检测频率和减少训练数据时性能更加稳定。

{"title":"Tool breakage monitoring driven by the real-time predicted spindle cutting torque using spindle servo signals","authors":"Yinghao Cheng , Yingguang Li , Guangxu Li , Xu Liu , Jinyu Xia , Changqing Liu , Xiaozhong Hao","doi":"10.1016/j.rcim.2024.102888","DOIUrl":"10.1016/j.rcim.2024.102888","url":null,"abstract":"<div><div>Monitoring tool breakage during computer numerical control machining is essential to ensure machining quality and equipment safety. In consideration of the low cost in long-term use and the non-invasiveness to workspace, using servo signals of machine tools to monitor tool breakage has been viewed as the solution that has great potential to be applied in real industry. However, because machine tool servo signals can only partially and indirectly reflect tool conditions, the accuracy and reliability of existing methods still need to be improved. To overcome this challenge, a novel two-step data-driven tool breakage monitoring method using spindle servo signals is proposed. Since spindle cutting torque is acknowledged as one of the most effective and reliable physical signals for detecting tool breakage, it is introduced as the key intermediate variable from spindle servo signals to tool conditions. The monitored spindle servo signals are used to predict the spindle cutting torque in real time based on a long short-term memory neural network, and then the predicted spindle cutting torque is used to detect tool breakage based on a one-dimensional convolutional neural network. The experimental results show that the proposed method can accurately predict the spindle cutting torque for normal tools and broken tools. Compared with the tool breakage monitoring methods that directly use spindle servo signals, the proposed method has higher detection accuracy and more reliable detection results, and the performance is more stable when increasing the detection frequency and decreasing training data.</div></div>","PeriodicalId":21452,"journal":{"name":"Robotics and Computer-integrated Manufacturing","volume":"92 ","pages":"Article 102888"},"PeriodicalIF":9.1,"publicationDate":"2024-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142433852","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