Automation in Construction最新文献

英文中文

Real-time robotic teleoperation for pavement pothole segmentation, quantification, and localization using multimodal sensing and efficient multi-scale attention-enhanced edge deep learning 利用多模态传感和高效的多尺度注意力增强边缘深度学习进行路面坑洼分割、量化和定位的实时机器人远程操作

IF 11.5 1区工程技术 Q1 CONSTRUCTION & BUILDING TECHNOLOGY

Automation in Construction

Pub Date : 2026-03-01 Epub Date: 2026-02-04 DOI: 10.1016/j.autcon.2026.106806

Xi Hu , Rayan H. Assaad

This paper proposes a robotic teleoperation pipeline to automate the segmentation, quantification, localization, and visualization of pavement potholes in real-time. The pipeline includes a new attention-based deep learning (DL) model and integrates a 4WD robot, teleoperation workstation, multimodal RGBD sensing fusion and point cloud processing on the edge, and interactive web application through cloud services. The DL model was developed by incorporating an efficient multi-scale attention (EMA) mechanism and transfer learning, which was trained and tested on a pavement dataset with 9472 images. The pipeline was validated through real-world field tests. The new EMA-based DL model yielded a 0.611 mAP50–95(B) and a 0.613 mAP50–95(M), outperforming the YOLOv9 baseline by 8.33% and 6.98%, respectively. The findings also showed that the proposed pipeline successfully automates pothole inspection and generates an interactive map, enabling remote access to the robot's trajectory and detailed pothole information, including pothole area, volume, average and maximum depth.

本文提出了一种机器人远程操作流水线，实现路面坑洼的实时分割、量化、定位和可视化。该管道包括一个新的基于注意力的深度学习（DL）模型，集成了一个四轮驱动机器人、远程操作工作站、多模态RGBD传感融合和边缘点云处理，以及通过云服务的交互式web应用程序。DL模型结合了高效的多尺度注意（EMA）机制和迁移学习，并在9472张图像的路面数据集上进行了训练和测试。该管道通过实际现场测试进行了验证。新的基于ema的DL模型产生了0.611 mAP50-95 (B)和0.613 mAP50-95 (M)，分别比YOLOv9基线高8.33%和6.98%。研究结果还表明，拟议的管道成功地实现了坑洼检测的自动化，并生成了交互式地图，可以远程访问机器人的轨迹和详细的坑洼信息，包括坑洼面积、体积、平均和最大深度。

{"title":"Real-time robotic teleoperation for pavement pothole segmentation, quantification, and localization using multimodal sensing and efficient multi-scale attention-enhanced edge deep learning","authors":"Xi Hu , Rayan H. Assaad","doi":"10.1016/j.autcon.2026.106806","DOIUrl":"10.1016/j.autcon.2026.106806","url":null,"abstract":"<div><div>This paper proposes a robotic teleoperation pipeline to automate the segmentation, quantification, localization, and visualization of pavement potholes in real-time. The pipeline includes a new attention-based deep learning (DL) model and integrates a 4WD robot, teleoperation workstation, multimodal RGBD sensing fusion and point cloud processing on the edge, and interactive web application through cloud services. The DL model was developed by incorporating an efficient multi-scale attention (EMA) mechanism and transfer learning, which was trained and tested on a pavement dataset with 9472 images. The pipeline was validated through real-world field tests. The new EMA-based DL model yielded a 0.611 mAP50–95(B) and a 0.613 mAP50–95(M), outperforming the YOLOv9 baseline by 8.33% and 6.98%, respectively. The findings also showed that the proposed pipeline successfully automates pothole inspection and generates an interactive map, enabling remote access to the robot's trajectory and detailed pothole information, including pothole area, volume, average and maximum depth.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"183 ","pages":"Article 106806"},"PeriodicalIF":11.5,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146185057","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Automated lightweight networks for multi-material bridge crack segmentation 多材料桥梁裂缝分割的自动化轻量化网络

IF 11.5 1区工程技术 Q1 CONSTRUCTION & BUILDING TECHNOLOGY

Automation in Construction

Pub Date : 2026-03-01 Epub Date: 2026-01-30 DOI: 10.1016/j.autcon.2026.106808

Mohammed Ameen Mohammed , Haijun Zhou , Jiaolei Zhang , Shikun Xu

Crack segmentation on concrete, steel, and asphalt surfaces remains challenging due to irregular crack patterns, low contrast, and noise interference, particularly in complex environments. Although deep neural network–based methods show promise, they often struggle to balance fine-grained feature extraction with contextual understanding. Moreover, no unified model effectively detects cracks across concrete, steel bridges, and asphalt pavements on bridge decks, while most existing models are too large for edge deployment. This paper introduces CrackSeg-GWD, a lightweight encoder–decoder model integrating Group Normalization, Weight-Standardized Convolutions, DropBlock regularization, and a Symmetric Unified Focal Loss to enhance stability, reduce overfitting, and handle class imbalance. With only 0.414 M parameters and 0.849 GFLOPs, it achieves high accuracy with low computational cost. Evaluated on five public datasets, SteelCrack, YCD, Crack500, DeepCrack, and Ozgenel, CrackSeg-GWD outperforms ten state-of-the-art models, achieving consistent gains across five metrics and confirming its suitability for real-time structural monitoring and construction automation.

由于不规则的裂缝模式、低对比度和噪声干扰，特别是在复杂的环境中，混凝土、钢铁和沥青表面的裂缝分割仍然具有挑战性。尽管基于深度神经网络的方法显示出前景，但它们往往难以平衡细粒度特征提取与上下文理解。此外，没有统一的模型可以有效地检测混凝土、钢桥和桥面沥青路面上的裂缝，而现有的大多数模型都太大，无法进行边缘部署。本文介绍了一种轻量级的编码器-解码器模型CrackSeg-GWD，该模型集成了群归一化、权重标准化卷积、DropBlock正则化和对称统一焦损失，以增强稳定性、减少过拟合和处理类不平衡。仅使用0.414 M参数和0.849 GFLOPs，以较低的计算成本实现了较高的精度。通过对SteelCrack、YCD、Crack500、DeepCrack和Ozgenel这5个公共数据集的评估，CrackSeg-GWD优于10个最先进的模型，在5个指标上取得了一致的收益，并证实了其在实时结构监测和施工自动化方面的适用性。

{"title":"Automated lightweight networks for multi-material bridge crack segmentation","authors":"Mohammed Ameen Mohammed , Haijun Zhou , Jiaolei Zhang , Shikun Xu","doi":"10.1016/j.autcon.2026.106808","DOIUrl":"10.1016/j.autcon.2026.106808","url":null,"abstract":"<div><div>Crack segmentation on concrete, steel, and asphalt surfaces remains challenging due to irregular crack patterns, low contrast, and noise interference, particularly in complex environments. Although deep neural network–based methods show promise, they often struggle to balance fine-grained feature extraction with contextual understanding. Moreover, no unified model effectively detects cracks across concrete, steel bridges, and asphalt pavements on bridge decks, while most existing models are too large for edge deployment. This paper introduces CrackSeg-GWD, a lightweight encoder–decoder model integrating Group Normalization, Weight-Standardized Convolutions, DropBlock regularization, and a Symmetric Unified Focal Loss to enhance stability, reduce overfitting, and handle class imbalance. With only 0.414 M parameters and 0.849 GFLOPs, it achieves high accuracy with low computational cost. Evaluated on five public datasets, SteelCrack, YCD, Crack500, DeepCrack, and Ozgenel, CrackSeg-GWD outperforms ten state-of-the-art models, achieving consistent gains across five metrics and confirming its suitability for real-time structural monitoring and construction automation.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"183 ","pages":"Article 106808"},"PeriodicalIF":11.5,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146074978","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Computer vision for infrastructure defect detection: Methods and trends 基础设施缺陷检测的计算机视觉：方法和趋势

IF 11.5 1区工程技术 Q1 CONSTRUCTION & BUILDING TECHNOLOGY

Automation in Construction

Pub Date : 2026-03-01 Epub Date: 2026-01-24 DOI: 10.1016/j.autcon.2026.106795

Yufei Zhang , Gang Li , Runjie Shen

Infrastructure defect detection is vital for public safety and sustainable societal development. In recent years, advances in computer vision have gradually promoted the intelligence and automation of infrastructure defect detection. This paper provides a comprehensive overview of research progress and emerging trends in computer vision-based detection of diverse defect types across multiple infrastructure scenarios, including datasets, evaluation metrics, and methods. A classification framework is introduced that centers on single and multiple visual modalities. The former includes traditional image processing, machine learning, and deep learning techniques, reflecting the evolution of the field. The latter focuses on data-level, feature-level, and decision-level fusion strategies, highlighting opportunities to improve detection performance with multiple visual modalities. Methods are further categorized according to their characteristics and model architectures. Finally, existing challenges are summarized, and promising research directions are outlined based on the strengths and limitations of current methods.

基础设施缺陷检测对公共安全和社会可持续发展至关重要。近年来，计算机视觉的进步逐步推动了基础设施缺陷检测的智能化和自动化。本文提供了研究进展的全面概述，以及跨多个基础设施场景的基于计算机视觉的各种缺陷类型检测的新兴趋势，包括数据集、评估度量和方法。介绍了一种以单视觉模态和多视觉模态为中心的分类框架。前者包括传统的图像处理、机器学习和深度学习技术，反映了该领域的发展。后者侧重于数据级、特征级和决策级融合策略，突出了通过多种视觉模式提高检测性能的机会。方法根据其特征和模型架构进一步分类。最后，根据现有方法的优势和局限性，总结了存在的挑战，并展望了未来的研究方向。

{"title":"Computer vision for infrastructure defect detection: Methods and trends","authors":"Yufei Zhang , Gang Li , Runjie Shen","doi":"10.1016/j.autcon.2026.106795","DOIUrl":"10.1016/j.autcon.2026.106795","url":null,"abstract":"<div><div>Infrastructure defect detection is vital for public safety and sustainable societal development. In recent years, advances in computer vision have gradually promoted the intelligence and automation of infrastructure defect detection. This paper provides a comprehensive overview of research progress and emerging trends in computer vision-based detection of diverse defect types across multiple infrastructure scenarios, including datasets, evaluation metrics, and methods. A classification framework is introduced that centers on single and multiple visual modalities. The former includes traditional image processing, machine learning, and deep learning techniques, reflecting the evolution of the field. The latter focuses on data-level, feature-level, and decision-level fusion strategies, highlighting opportunities to improve detection performance with multiple visual modalities. Methods are further categorized according to their characteristics and model architectures. Finally, existing challenges are summarized, and promising research directions are outlined based on the strengths and limitations of current methods.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"183 ","pages":"Article 106795"},"PeriodicalIF":11.5,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146033310","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Edge device-based vibration signal processing and convolutional neural networks for mining dumper activity recognition 基于边缘设备的振动信号处理和卷积神经网络的矿用自卸车活动识别

IF 11.5 1区工程技术 Q1 CONSTRUCTION & BUILDING TECHNOLOGY

Automation in Construction

Pub Date : 2026-03-01 Epub Date: 2026-01-19 DOI: 10.1016/j.autcon.2026.106785

Nagesh Dewangan, Amiya Ranjan Mohanty

Previous research on mining dumper activity recognition has rarely explored edge devices for classifying processed vibration signals from deployed Convolutional Neural Network (CNN) models. Most studies have relied on remote or cloud-based platforms, limiting applicability in mines due to unreliable connectivity. This paper introduces an on-device classification approach for dumper activities from processed vibration signals by deploying trained CNN models on edge devices. Vibration signals collected were processed using signal processing methods to extract distinct features for classification and validated through SHAP 3D surface visualization. Among tested models, the combination of ResNet50 with DWT-GT achieved optimal performance, delivering 99.23% accuracy with low computational complexity. Deployment on resource-constrained devices demonstrated feasibility of edge-based computation, where BeagleBone AI-64 achieved 67.46% lower CPU time. These findings establish the feasibility of edge devices for real-time dumper activity recognition, eliminating dependency on external platforms and enhancing operational efficiency in mining environments.

以往关于矿用自卸车活动识别的研究很少探索利用卷积神经网络（CNN）模型对处理后的振动信号进行分类的边缘设备。大多数研究都依赖于远程或基于云的平台，由于连接不可靠，限制了在矿山中的适用性。本文介绍了一种通过在边缘设备上部署训练好的CNN模型，从处理过的振动信号中对翻车机活动进行设备上分类的方法。采用信号处理方法对采集到的振动信号进行处理，提取明显特征进行分类，并通过SHAP三维表面可视化进行验证。在测试的模型中，ResNet50与DWT-GT的组合获得了最佳性能，准确率达到99.23%，计算复杂度较低。在资源受限设备上的部署证明了边缘计算的可行性，其中BeagleBone AI-64的CPU时间降低了67.46%。这些发现确定了边缘设备用于实时卸料器活动识别的可行性，消除了对外部平台的依赖，提高了采矿环境中的操作效率。

{"title":"Edge device-based vibration signal processing and convolutional neural networks for mining dumper activity recognition","authors":"Nagesh Dewangan, Amiya Ranjan Mohanty","doi":"10.1016/j.autcon.2026.106785","DOIUrl":"10.1016/j.autcon.2026.106785","url":null,"abstract":"<div><div>Previous research on mining dumper activity recognition has rarely explored edge devices for classifying processed vibration signals from deployed Convolutional Neural Network (CNN) models. Most studies have relied on remote or cloud-based platforms, limiting applicability in mines due to unreliable connectivity. This paper introduces an on-device classification approach for dumper activities from processed vibration signals by deploying trained CNN models on edge devices. Vibration signals collected were processed using signal processing methods to extract distinct features for classification and validated through SHAP 3D surface visualization. Among tested models, the combination of ResNet50 with DWT-GT achieved optimal performance, delivering 99.23% accuracy with low computational complexity. Deployment on resource-constrained devices demonstrated feasibility of edge-based computation, where BeagleBone AI-64 achieved 67.46% lower CPU time. These findings establish the feasibility of edge devices for real-time dumper activity recognition, eliminating dependency on external platforms and enhancing operational efficiency in mining environments.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"183 ","pages":"Article 106785"},"PeriodicalIF":11.5,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146000935","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Constructability-aware Physics-Informed Graph Neural Networks for surrogate-assisted optimization design of rebar in concrete beams 基于可构造性感知的物理信息图神经网络在混凝土梁中钢筋的代理辅助优化设计

IF 11.5 1区工程技术 Q1 CONSTRUCTION & BUILDING TECHNOLOGY

Automation in Construction

Pub Date : 2026-02-01 Epub Date: 2026-01-07 DOI: 10.1016/j.autcon.2026.106760

Luis F. Verduzco , Jack C.P. Cheng , Mingkai Li , Lufeng Wang

Constructability-based optimization design of reinforcing bar (rebar) in Reinforced Concrete (RC) structures is crucial for more sustainable practices in the construction industry. To make the process time-efficient, the use of surrogate models is necessary, especially with Graph Neural Networks (GNNs). However, the adoption of GNNs alone can be limited for large RC buildings, due to their characterization as data-hungry models. In this context, Physics-Informed Neural Networks become relevant. Their implementation, however, remains unexploited for this task, where constructability constraints are as preponderant as the physics behind. This paper presents a Constructability-Aware Physics-Informed Graph Neural Network (PIGNN) for surrogate-assisted optimization design of rebar in concrete beams (CPyRO-GraphNet-Beams). Its testing and application for fixed-end supported beams is presented, as a comparison with Plain GNNs. It is demonstrated that CPyRO-GraphNet-Beams outperforms Plain GNNs, highlighting its greater capability to learn constructable features from datasets, enhancing, in turn, more practical and sustainable optimum rebar designs.

基于可施工性的钢筋混凝土（RC）结构中钢筋的优化设计对于建筑行业的可持续实践至关重要。为了提高过程的时间效率，使用代理模型是必要的，特别是对于图神经网络（gnn）。然而，对于大型RC建筑来说，单独采用gnn可能会受到限制，因为它们的特征是数据饥渴模型。在这种情况下，物理信息神经网络变得相关。然而，它们的实现在这个任务中仍然没有被利用，在这个任务中，可构造性约束和背后的物理一样占主导地位。提出了一种可构造性感知的物理信息图神经网络（PIGNN），用于混凝土梁中钢筋的代理辅助优化设计（cpor - graphnet - beams）。介绍了它在固定端支承梁中的测试和应用，并与普通gnn进行了比较。研究表明，cpro - graphnet - beams优于Plain GNNs，突出了其从数据集中学习可构造特征的更强能力，从而增强了更实用和可持续的最佳螺杆钢设计。

{"title":"Constructability-aware Physics-Informed Graph Neural Networks for surrogate-assisted optimization design of rebar in concrete beams","authors":"Luis F. Verduzco , Jack C.P. Cheng , Mingkai Li , Lufeng Wang","doi":"10.1016/j.autcon.2026.106760","DOIUrl":"10.1016/j.autcon.2026.106760","url":null,"abstract":"<div><div>Constructability-based optimization design of reinforcing bar (rebar) in Reinforced Concrete (RC) structures is crucial for more sustainable practices in the construction industry. To make the process time-efficient, the use of surrogate models is necessary, especially with Graph Neural Networks (GNNs). However, the adoption of GNNs alone can be limited for large RC buildings, due to their characterization as data-hungry models. In this context, Physics-Informed Neural Networks become relevant. Their implementation, however, remains unexploited for this task, where constructability constraints are as preponderant as the physics behind. This paper presents a Constructability-Aware Physics-Informed Graph Neural Network (PIGNN) for surrogate-assisted optimization design of rebar in concrete beams (CPyRO-GraphNet-Beams). Its testing and application for fixed-end supported beams is presented, as a comparison with Plain GNNs. It is demonstrated that CPyRO-GraphNet-Beams outperforms Plain GNNs, highlighting its greater capability to learn constructable features from datasets, enhancing, in turn, more practical and sustainable optimum rebar designs.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"182 ","pages":"Article 106760"},"PeriodicalIF":11.5,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145920965","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Leveraging large language models for BIM-based automated compliance checking 利用大型语言模型进行基于bim的自动遵从性检查

IF 11.5 1区工程技术 Q1 CONSTRUCTION & BUILDING TECHNOLOGY

Automation in Construction

Pub Date : 2026-02-01 Epub Date: 2025-12-05 DOI: 10.1016/j.autcon.2025.106707

Odin Iversen, Lizhen Huang

Current methods of checking regulatory compliance in the architecture, engineering, construction, and operations (AECO) industry are mostly manual, time consuming and error prone. This paper, using design science research (DSR), proposes an artifact that leverages a large language model (LLM) for automated compliance checking (ACC) to directly interpret regulations, extract BIM data, execute checks, and generate detailed reports. For rule interpretation, the artifact achieves high F1-scores (97% for classification, 100% for dependency identification). For building model preparation, it correctly selected data extraction tools in 97% of cases. In rule execution, it demonstrated 97,7% accuracy and significantly outperformed a naive baseline, which highlights the need for a structured framework. Finally, the artifact generated detailed reports that included the LLM’s reasoning. The key finding is that an LLM-based reasoning engine enables a holistic approach that overcomes the manual rule digitization bottleneck in traditional ACC systems.

在架构、工程、构造和操作（AECO）行业中，当前检查法规遵从性的方法大多是手动的、耗时且容易出错的。本文利用设计科学研究（DSR），提出了一种工件，该工件利用大型语言模型（LLM）进行自动合规性检查（ACC），直接解释法规，提取BIM数据，执行检查并生成详细报告。对于规则解释，工件达到了很高的f1分数（97%用于分类，100%用于依赖标识）。对于构建模型的准备，它在97%的情况下正确选择了数据提取工具。在规则执行中，它显示了97,7%的准确性，并且显著优于朴素基线，这突出了对结构化框架的需求。最后，工件生成了包含LLM推理的详细报告。关键发现是，基于llm的推理引擎实现了一种整体方法，克服了传统ACC系统中手动规则数字化的瓶颈。

引用次数: 0

Adaptive reinforcement learning algorithm for real-time energy optimization in building digital twins with heterogeneous IoT sensor networks 基于异构物联网传感器网络的数字孪生建筑实时能量优化自适应强化学习算法

IF 11.5 1区工程技术 Q1 CONSTRUCTION & BUILDING TECHNOLOGY

Automation in Construction

Pub Date : 2026-02-01 Epub Date: 2025-12-13 DOI: 10.1016/j.autcon.2025.106714

Manea Almatared , Maher Abuhussain , Zahra Andleeb , Faizah Mohammed Bashir

Building operations contribute significantly to global carbon emissions, yet existing digital twins rely on static strategies that fail to adapt to dynamic conditions. This paper investigates whether an adaptive reinforcement learning framework with hierarchical transfer learning can optimize real-time energy use and comfort across heterogeneous buildings without extensive retraining. The paper develops and validates a proximal policy optimization controller integrated with a digital twin and 348 IoT sensors across three commercial buildings in a 14-month randomized crossover trial. The approach achieves 23.7 % and 14.9 % energy reductions compared to rule-based control and model predictive control, respectively, while maintaining 94.7 % comfort satisfaction and demonstrating 78 % transfer learning efficiency. These findings provide facility managers and grid operators with a scalable, hardware-validated approach that reduces operational costs and stabilizes demand response without compromising occupant comfort. Future work extends this hierarchical transfer framework to mixed-mode ventilation and district-level energy coordination to further enhance grid interactivity.

建筑运营对全球碳排放的贡献很大，但现有的数字孪生依赖于静态策略，无法适应动态条件。本文研究了具有分层迁移学习的自适应强化学习框架是否可以在不进行大量再培训的情况下优化异构建筑的实时能源使用和舒适度。本文在为期14个月的随机交叉试验中，在三座商业建筑中开发并验证了集成了数字孪生和348个物联网传感器的近端策略优化控制器。与基于规则的控制和模型预测控制相比，该方法分别降低了23.7%和14.9%的能量，同时保持了94.7%的舒适性满意度和78%的迁移学习效率。这些发现为设施管理人员和电网运营商提供了一种可扩展的、经过硬件验证的方法，可以降低运营成本，稳定需求响应，同时不影响居住者的舒适度。未来的工作将这种分层转移框架扩展到混合模式通风和区域一级的能源协调，以进一步增强电网的交互性。

{"title":"Adaptive reinforcement learning algorithm for real-time energy optimization in building digital twins with heterogeneous IoT sensor networks","authors":"Manea Almatared , Maher Abuhussain , Zahra Andleeb , Faizah Mohammed Bashir","doi":"10.1016/j.autcon.2025.106714","DOIUrl":"10.1016/j.autcon.2025.106714","url":null,"abstract":"<div><div>Building operations contribute significantly to global carbon emissions, yet existing digital twins rely on static strategies that fail to adapt to dynamic conditions. This paper investigates whether an adaptive reinforcement learning framework with hierarchical transfer learning can optimize real-time energy use and comfort across heterogeneous buildings without extensive retraining. The paper develops and validates a proximal policy optimization controller integrated with a digital twin and 348 IoT sensors across three commercial buildings in a 14-month randomized crossover trial. The approach achieves 23.7 % and 14.9 % energy reductions compared to rule-based control and model predictive control, respectively, while maintaining 94.7 % comfort satisfaction and demonstrating 78 % transfer learning efficiency. These findings provide facility managers and grid operators with a scalable, hardware-validated approach that reduces operational costs and stabilizes demand response without compromising occupant comfort. Future work extends this hierarchical transfer framework to mixed-mode ventilation and district-level energy coordination to further enhance grid interactivity.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"182 ","pages":"Article 106714"},"PeriodicalIF":11.5,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145759499","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Integrating metaheuristic optimization algorithms with random forest to predict waste generation in construction and demolition projects 结合随机森林的元启发式优化算法预测建筑和拆除工程中的废物产生

IF 11.5 1区工程技术 Q1 CONSTRUCTION & BUILDING TECHNOLOGY

Automation in Construction

Pub Date : 2026-02-01 Epub Date: 2025-12-19 DOI: 10.1016/j.autcon.2025.106732

Ruba Awad , Cenk Budayan , Idil Calik , Aslı Pelin Gurgun , Kerim Koc

The construction sector is a significant source of global waste, making accurate and proactive prediction of Construction and Demolition Waste (C&DW) essential for sustainable resource management and circular economy efforts. However, estimating C&DW at the project level remains a major challenge. This paper investigates whether C&DW prediction accuracy can be enhanced by integrating the Random Forest (RF) model with two metaheuristic optimization algorithms: the Archimedes Optimization Algorithm (AOA) and Grey Wolf Optimization (GWO). Based on data from 200 real-world projects in Palestine, the GWO-RF model achieved the highest predictive accuracy using only four input variables: project type, start date, building type, and number of floors. To ensure model transparency, Shapley Additive Explanations (SHAP) analysis confirmed that project type and the number of floors were the most influential parameters. This study thus provides a practical, robust, and highly accurate model to support effective waste management strategies in the construction industry.

建筑业是全球废物的重要来源，因此对建筑和拆除废物（C&；DW）进行准确和主动的预测对于可持续资源管理和循环经济的努力至关重要。然而，在项目级别估计C&；DW仍然是一个主要的挑战。本文研究了随机森林（RF）模型与阿基米德优化算法（AOA）和灰狼优化算法（GWO）两种元启发式优化算法相结合，能否提高C&；DW的预测精度。基于巴勒斯坦200个实际项目的数据，GWO-RF模型仅使用四个输入变量（项目类型、开始日期、建筑类型和楼层数）就实现了最高的预测精度。为了确保模型的透明度，Shapley加性解释（SHAP）分析证实，项目类型和楼层数量是最具影响力的参数。因此，本研究提供了一个实用、稳健和高度准确的模型，以支持建筑行业有效的废物管理策略。

{"title":"Integrating metaheuristic optimization algorithms with random forest to predict waste generation in construction and demolition projects","authors":"Ruba Awad , Cenk Budayan , Idil Calik , Aslı Pelin Gurgun , Kerim Koc","doi":"10.1016/j.autcon.2025.106732","DOIUrl":"10.1016/j.autcon.2025.106732","url":null,"abstract":"<div><div>The construction sector is a significant source of global waste, making accurate and proactive prediction of Construction and Demolition Waste (C&DW) essential for sustainable resource management and circular economy efforts. However, estimating C&DW at the project level remains a major challenge. This paper investigates whether C&DW prediction accuracy can be enhanced by integrating the Random Forest (RF) model with two metaheuristic optimization algorithms: the Archimedes Optimization Algorithm (AOA) and Grey Wolf Optimization (GWO). Based on data from 200 real-world projects in Palestine, the GWO-RF model achieved the highest predictive accuracy using only four input variables: project type, start date, building type, and number of floors. To ensure model transparency, Shapley Additive Explanations (SHAP) analysis confirmed that project type and the number of floors were the most influential parameters. This study thus provides a practical, robust, and highly accurate model to support effective waste management strategies in the construction industry.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"182 ","pages":"Article 106732"},"PeriodicalIF":11.5,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145786090","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Domain-adaptive instance segmentation for far-field object monitoring using SAM-based weak supervision and noisy student self-training 基于sam的弱监督和噪声学生自训练的远场目标监测领域自适应实例分割

IF 11.5 1区工程技术 Q1 CONSTRUCTION & BUILDING TECHNOLOGY

Automation in Construction

Pub Date : 2026-02-01 Epub Date: 2026-01-13 DOI: 10.1016/j.autcon.2026.106772

Minkyu Koo , Taegeon Kim , Minhyun Lee , Kinam Kim , Hongjo Kim

Automating construction site monitoring through deep learning–based segmentation presents challenges due to the high cost of pixel-wise annotations. This paper introduces a weakly and self-supervised learning framework that enhances segmentation accuracy while reducing annotation burden. Human-annotated bounding-box ground truth is used as prompts for the Segment Anything Model (SAM) to generate high-quality polygon mask labels, which are further refined through self-training. Compared to fully supervised learning models, the framework integrates Transfer Learning, Pseudo-Label Refinement, and the Noisy Student technique, improving mask mean Average Precision (Mask mAP) by 3–63% across seven target domains and achieving a Mask mAP of 72.27%. The approach also outperforms existing weakly supervised techniques, including BoxSnake and BoxTeacher, by 18% and 25.95%, respectively, and exceeds the performance of point-based methods such as PointWSSIS by 48.78%.

由于像素级标注的高成本，通过基于深度学习的分割自动化施工现场监控提出了挑战。本文引入了一种弱自监督学习框架，在降低标注负担的同时提高了分割精度。将人类标注的边界框地面真值作为SAM （Segment Anything Model）的提示符，生成高质量的多边形掩码标签，并通过自我训练进一步细化。与完全监督学习模型相比，该框架集成了迁移学习、伪标签细化和噪声学生技术，在7个目标域将mask mean Average Precision （mask mAP）提高了3-63%，实现了72.27%的mask mAP。该方法也比现有的弱监督技术（包括BoxSnake和BoxTeacher）分别高出18%和25.95%，并且比基于点的方法（如PointWSSIS）的性能高出48.78%。

引用次数: 0

Structural defect segmentation using a semi-supervised algorithm integrating YOLO and the segment anything model 结合YOLO和任意分割模型的半监督结构缺陷分割算法

IF 11.5 1区工程技术 Q1 CONSTRUCTION & BUILDING TECHNOLOGY

Automation in Construction

Pub Date : 2026-02-01 Epub Date: 2025-12-11 DOI: 10.1016/j.autcon.2025.106709

Gauthami Vijayakumar Kuttuva, Prawin J

Deep learning–based defect segmentation enables automated detection, classification, localisation, and quantification of structural defects. Although supervised segmentation methods yield strong performance, they require extensive pixel-level annotations and struggle with data scarcity, class imbalance, thin crack segmentation, and generalisation across diverse conditions. This paper proposes a semi-supervised two-stage segmentation framework that integrates a high-accuracy object detection model (Stage I: localisation using YOLO11/Oriented YOLO11) with zero-shot unsupervised segmentation (Stage II: pixel-level mapping using Segment Anything Model with box prompts). The method requires only object-detection labels, eliminating the need for dedicated segmentation annotations. Experiments on benchmark datasets involving steel, masonry, concrete cracks, spalling, and corrosion demonstrate that the hybrid Oriented YOLO11 and SAM model achieves state-of-the-art performance, with an average Dice score of 0.7, comparable to those of fully supervised models. The proposed framework offers real-time performance, strong generalisation, and high scalability, making it a promising solution for robust structural defect segmentation.

基于深度学习的缺陷分割实现了结构缺陷的自动检测、分类、定位和量化。尽管监督分割方法产生了强大的性能，但它们需要大量的像素级注释，并与数据稀缺性、类不平衡、细裂纹分割和不同条件下的泛化作斗争。本文提出了一种半监督两阶段分割框架，该框架集成了高精度目标检测模型（第一阶段：使用YOLO11/Oriented YOLO11进行定位）和零镜头无监督分割（第二阶段：使用带有框提示的Segment Anything model进行像素级映射）。该方法只需要对象检测标签，不需要专门的分割注释。在涉及钢铁、砌体、混凝土裂缝、剥落和腐蚀的基准数据集上的实验表明，混合的面向YOLO11和SAM模型达到了最先进的性能，平均Dice得分为0.7，与完全监督模型相当。该框架具有实时性好、泛化能力强、可扩展性强等特点，是鲁棒结构缺陷分割的理想解决方案。

{"title":"Structural defect segmentation using a semi-supervised algorithm integrating YOLO and the segment anything model","authors":"Gauthami Vijayakumar Kuttuva, Prawin J","doi":"10.1016/j.autcon.2025.106709","DOIUrl":"10.1016/j.autcon.2025.106709","url":null,"abstract":"<div><div>Deep learning–based defect segmentation enables automated detection, classification, localisation, and quantification of structural defects. Although supervised segmentation methods yield strong performance, they require extensive pixel-level annotations and struggle with data scarcity, class imbalance, thin crack segmentation, and generalisation across diverse conditions. This paper proposes a semi-supervised two-stage segmentation framework that integrates a high-accuracy object detection model (Stage I: localisation using YOLO11/Oriented YOLO11) with zero-shot unsupervised segmentation (Stage II: pixel-level mapping using Segment Anything Model with box prompts). The method requires only object-detection labels, eliminating the need for dedicated segmentation annotations. Experiments on benchmark datasets involving steel, masonry, concrete cracks, spalling, and corrosion demonstrate that the hybrid Oriented YOLO11 and SAM model achieves state-of-the-art performance, with an average Dice score of 0.7, comparable to those of fully supervised models. The proposed framework offers real-time performance, strong generalisation, and high scalability, making it a promising solution for robust structural defect segmentation.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"182 ","pages":"Article 106709"},"PeriodicalIF":11.5,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145734739","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

首页上一页

下一页尾页

类型

全部化学•材料生命科学医学物理工程技术环境•农林材料科学地球科学法学管理学化学环境科学与生态学计算机科学教育学经济学农林科学人文科学生物学数学物理与天体物理心理学综合性期刊其他工业工程理学历史学农学文学信息工程

数据库

全部 ACS Publications Elsevier ieeexplore Springer The Royal Society of Chemistry Wiley

期刊

Automation in Construction

全部 Acc. Chem. Res. ACS Applied Bio Materials ACS Appl. Electron. Mater. ACS Appl. Energy Mater. ACS Appl. Mater. Interfaces ACS Appl. Nano Mater. ACS Appl. Polym. Mater. ACS BIOMATER-SCI ENG ACS Catal. ACS Cent. Sci. ACS Chem. Biol. ACS Chemical Health & Safety ACS Chem. Neurosci. ACS Comb. Sci. ACS Earth Space Chem. ACS Energy Lett. ACS Infect. Dis. ACS Macro Lett. ACS Mater. Lett. ACS Med. Chem. Lett. ACS Nano ACS Omega ACS Photonics ACS Sens. ACS Sustainable Chem. Eng. ACS Synth. Biol. Anal. Chem. BIOCHEMISTRY-US Bioconjugate Chem. BIOMACROMOLECULES Chem. Res. Toxicol. Chem. Rev. Chem. Mater. CRYST GROWTH DES ENERG FUEL Environ. Sci. Technol. Environ. Sci. Technol. Lett. Eur. J. Inorg. Chem. IND ENG CHEM RES Inorg. Chem. J. Agric. Food. Chem. J. Chem. Eng. Data J. Chem. Educ. J. Chem. Inf. Model. J. Chem. Theory Comput. J. Med. Chem. J. Nat. Prod. J PROTEOME RES J. Am. Chem. Soc. LANGMUIR MACROMOLECULES Mol. Pharmaceutics Nano Lett. Org. Lett. ORG PROCESS RES DEV ORGANOMETALLICS J. Org. Chem. J. Phys. Chem. J. Phys. Chem. A J. Phys. Chem. B J. Phys. Chem. C J. Phys. Chem. Lett. Analyst Anal. Methods Biomater. Sci. Catal. Sci. Technol. Chem. Commun. Chem. Soc. Rev. CHEM EDUC RES PRACT CRYSTENGCOMM Dalton Trans. Energy Environ. Sci. ENVIRON SCI-NANO ENVIRON SCI-PROC IMP ENVIRON SCI-WAT RES Faraday Discuss. Food Funct. Green Chem. Inorg. Chem. Front. Integr. Biol. J. Anal. At. Spectrom. J. Mater. Chem. A J. Mater. Chem. B J. Mater. Chem. C Lab Chip Mater. Chem. Front. Mater. Horiz. MEDCHEMCOMM Metallomics Mol. Biosyst. Mol. Syst. Des. Eng. Nanoscale Nanoscale Horiz. Nat. Prod. Rep. New J. Chem. Org. Biomol. Chem. Org. Chem. Front. PHOTOCH PHOTOBIO SCI PCCP Polym. Chem.

﹀