Pub Date : 2024-12-17DOI: 10.1016/j.autcon.2024.105901
Limao Zhang, Jiaqi Wang, Zhuang Xia, Xieqing Song
The overlapping construction of large-diameter tunnels is inevitable, but the construction control faces great challenges due to the complexity of underground environments. A generative design method for large-diameter tunnels under existing metro lines based on physic-guided deep learning is proposed, aiming at optimizing tunnel layouts from a physical perspective to ensure effective construction control. A topology-optimized model dataset considering soil uncertainties is fed into a physics-guided Wasserstein generative adversarial network (PGWGAN) for training, producing numerous physically consistent schemes. The optimal scheme is selected using the multiple-attribute decision-making (MADM) method under multi-working conditions. A case study on large-diameter tunnel construction demonstrates the method's feasibility, showing that it meets the safety requirements across various conditions and achieves significant improvement. This paper contributes a physics-guided generative design method for large-diameter tunnel overlapping construction. It accounts for multiple working conditions and includes an evaluation module that integrates parametric finite element analysis (FEA) with multi-attribute evaluation.
{"title":"Physics-guided deep learning for generative design of large-diameter tunnels under existing metro lines","authors":"Limao Zhang, Jiaqi Wang, Zhuang Xia, Xieqing Song","doi":"10.1016/j.autcon.2024.105901","DOIUrl":"https://doi.org/10.1016/j.autcon.2024.105901","url":null,"abstract":"The overlapping construction of large-diameter tunnels is inevitable, but the construction control faces great challenges due to the complexity of underground environments. A generative design method for large-diameter tunnels under existing metro lines based on physic-guided deep learning is proposed, aiming at optimizing tunnel layouts from a physical perspective to ensure effective construction control. A topology-optimized model dataset considering soil uncertainties is fed into a physics-guided Wasserstein generative adversarial network (PGWGAN) for training, producing numerous physically consistent schemes. The optimal scheme is selected using the multiple-attribute decision-making (MADM) method under multi-working conditions. A case study on large-diameter tunnel construction demonstrates the method's feasibility, showing that it meets the safety requirements across various conditions and achieves significant improvement. This paper contributes a physics-guided generative design method for large-diameter tunnel overlapping construction. It accounts for multiple working conditions and includes an evaluation module that integrates parametric finite element analysis (FEA) with multi-attribute evaluation.","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"83 1","pages":""},"PeriodicalIF":10.3,"publicationDate":"2024-12-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142867673","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-12-16DOI: 10.1016/j.autcon.2024.105937
Zhong Wang, Vicente A. González, Qipei Mei, Gaang Lee
This paper examines the underutilization of sensors in the construction industry despite their significant potential for improving performance. A systematic review was conducted on research published between 2004 and 2024, identifying 11 key barriers such as the need for advanced skill sets and user-centric design, lack of standardized practices, and challenges in data networks and management. The study applied both quantitative descriptive analysis and qualitative content analysis to explore these barriers across five stages of sensor adoption. A total of 63 articles were thoroughly reviewed to identify thematic patterns and chronological trends. The findings highlight critical areas that require attention, including the development of standardized protocols, enhancing data-driven decision-making with advanced analytics, and fostering industry-wide training programs. Additionally, leveraging Lean Construction 4.0 principles is proposed to address these challenges. The insights from this research aim to support the construction industry in integrating sensor technologies more effectively, leading to greater efficiency and improved performance.
{"title":"Sensor adoption in the construction industry: Barriers, opportunities, and strategies","authors":"Zhong Wang, Vicente A. González, Qipei Mei, Gaang Lee","doi":"10.1016/j.autcon.2024.105937","DOIUrl":"https://doi.org/10.1016/j.autcon.2024.105937","url":null,"abstract":"This paper examines the underutilization of sensors in the construction industry despite their significant potential for improving performance. A systematic review was conducted on research published between 2004 and 2024, identifying 11 key barriers such as the need for advanced skill sets and user-centric design, lack of standardized practices, and challenges in data networks and management. The study applied both quantitative descriptive analysis and qualitative content analysis to explore these barriers across five stages of sensor adoption. A total of 63 articles were thoroughly reviewed to identify thematic patterns and chronological trends. The findings highlight critical areas that require attention, including the development of standardized protocols, enhancing data-driven decision-making with advanced analytics, and fostering industry-wide training programs. Additionally, leveraging Lean Construction 4.0 principles is proposed to address these challenges. The insights from this research aim to support the construction industry in integrating sensor technologies more effectively, leading to greater efficiency and improved performance.","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"113 1","pages":""},"PeriodicalIF":10.3,"publicationDate":"2024-12-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142867674","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-12-16DOI: 10.1016/j.autcon.2024.105925
Hongyu Zhao, Junbo Sun, Xiangyu Wang, Yufei Wang, Yang Su, Jun Wang, Li Wang
Defects and anomalies during the 3D concrete printing (3DCP) process significantly affect final construction quality. This paper proposes a real-time, high-accuracy method for monitoring defects in the printing process using a transformer-based detector. Despite limited data availability, deep learning-based data augmentation and image processing techniques were employed to enable effective training of this complex transformer model. A range of enhancement strategies was applied to the RT-DETR, resulting in remarkable improvements, including a mAP50 of 98.1 %, mAP50–95 of 68.0 %, and a computation speed of 72 FPS. The enhanced RT-DETR outperformed state-of-the-art detectors such as YOLOv8 and YOLOv7 in detecting defects in 3DCP. Furthermore, the improved RT-DETR was used to analyze the relationships between defect count, size, and printer parameters, providing guidance for operators to fine-tune printer settings and promptly address defects. This monitoring method reduces material waste and minimizes the risk of structural collapse during the printing process.
{"title":"Real-time and high-accuracy defect monitoring for 3D concrete printing using transformer networks","authors":"Hongyu Zhao, Junbo Sun, Xiangyu Wang, Yufei Wang, Yang Su, Jun Wang, Li Wang","doi":"10.1016/j.autcon.2024.105925","DOIUrl":"https://doi.org/10.1016/j.autcon.2024.105925","url":null,"abstract":"Defects and anomalies during the 3D concrete printing (3DCP) process significantly affect final construction quality. This paper proposes a real-time, high-accuracy method for monitoring defects in the printing process using a transformer-based detector. Despite limited data availability, deep learning-based data augmentation and image processing techniques were employed to enable effective training of this complex transformer model. A range of enhancement strategies was applied to the RT-DETR, resulting in remarkable improvements, including a mAP50 of 98.1 %, mAP50–95 of 68.0 %, and a computation speed of 72 FPS. The enhanced RT-DETR outperformed state-of-the-art detectors such as YOLOv8 and YOLOv7 in detecting defects in 3DCP. Furthermore, the improved RT-DETR was used to analyze the relationships between defect count, size, and printer parameters, providing guidance for operators to fine-tune printer settings and promptly address defects. This monitoring method reduces material waste and minimizes the risk of structural collapse during the printing process.","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"14 1","pages":""},"PeriodicalIF":10.3,"publicationDate":"2024-12-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142867683","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-12-14DOI: 10.1016/j.autcon.2024.105889
Muhammad Muddassir, Tarek Zayed, Ali Hassan Ali, Mohamed Elrifaee, Sulemana Fatoama Abdulai, Tong Yang, Amr Eldemiry
Tower cranes play a vital role in modern construction for transporting material, yet the persisting issue of crane-related accidents, often attributable to human error, underscores the urgent need for automated crane operations to enhance safety on construction sites. Despite active research in this area, a gap exists in systematically examining and categorising advancements in tower crane automation and identifying key trends and limitations. This paper aims to address this gap by employing a mixed-methods approach, encompassing scientometric and systematic analyses. The scientometric analysis sheds light on key researchers, institutions, journals, and global research networks. Also, the systematic analysis delves into four primary research areas: crane operations, motion control, layout planning, and transport path optimisation. This paper identifies critical knowledge gaps and limitations in tower crane automation, suggests future research directions, and offers industry insights into current methodologies and global trends.
{"title":"Automation in tower cranes over the past two decades (2003–2024)","authors":"Muhammad Muddassir, Tarek Zayed, Ali Hassan Ali, Mohamed Elrifaee, Sulemana Fatoama Abdulai, Tong Yang, Amr Eldemiry","doi":"10.1016/j.autcon.2024.105889","DOIUrl":"https://doi.org/10.1016/j.autcon.2024.105889","url":null,"abstract":"Tower cranes play a vital role in modern construction for transporting material, yet the persisting issue of crane-related accidents, often attributable to human error, underscores the urgent need for automated crane operations to enhance safety on construction sites. Despite active research in this area, a gap exists in systematically examining and categorising advancements in tower crane automation and identifying key trends and limitations. This paper aims to address this gap by employing a mixed-methods approach, encompassing scientometric and systematic analyses. The scientometric analysis sheds light on key researchers, institutions, journals, and global research networks. Also, the systematic analysis delves into four primary research areas: crane operations, motion control, layout planning, and transport path optimisation. This paper identifies critical knowledge gaps and limitations in tower crane automation, suggests future research directions, and offers industry insights into current methodologies and global trends.","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"12 Suppl 1 1","pages":""},"PeriodicalIF":10.3,"publicationDate":"2024-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142867680","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}
3D laser scanning can serve the geometric deformation detection of steel structures. However, the process of handling large-scale point clouds remains labor-intensive and time-consuming. This paper presents an automated approach to extracting the precise axes from point clouds and updating the associated BIM model for steel structures. The strategy involves the initial geometry extraction from IFC files and instance segmentation through the reference point cloud simplification and index rules. Then the axes of all components with different sections are detected through the corresponding standard sections and genetic algorithm. Lastly, the geometric information for each component in the BIM is updated by modifying the IFC file. The method is implemented on a steel framing comprising 218 components, indicating that the workflow works effectively with noise and occlusion. The difference in average distances from 218 components to the scanned point cloud is reduced from 17.50 mm before updating to 4.00 mm after updating.
{"title":"Scan vs. BIM: Automated geometry detection and BIM updating of steel framing through laser scanning","authors":"Siwei Lin, Liping Duan, Bin Jiang, Jiming Liu, Haoyu Guo, Jincheng Zhao","doi":"10.1016/j.autcon.2024.105931","DOIUrl":"https://doi.org/10.1016/j.autcon.2024.105931","url":null,"abstract":"3D laser scanning can serve the geometric deformation detection of steel structures. However, the process of handling large-scale point clouds remains labor-intensive and time-consuming. This paper presents an automated approach to extracting the precise axes from point clouds and updating the associated BIM model for steel structures. The strategy involves the initial geometry extraction from IFC files and instance segmentation through the reference point cloud simplification and index rules. Then the axes of all components with different sections are detected through the corresponding standard sections and genetic algorithm. Lastly, the geometric information for each component in the BIM is updated by modifying the IFC file. The method is implemented on a steel framing comprising 218 components, indicating that the workflow works effectively with noise and occlusion. The difference in average distances from 218 components to the scanned point cloud is reduced from 17.50 mm before updating to 4.00 mm after updating.","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"38 1","pages":""},"PeriodicalIF":10.3,"publicationDate":"2024-12-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142816503","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-12-13DOI: 10.1016/j.autcon.2024.105898
Junjie Gong, Jian Chen, Dengsheng Cai, Wei Wei, Yu Long
Trajectory tracking control is pivotal for achieving autonomous operation in hydraulic excavators. This paper proposes a robust control scheme, merging passivity-based and impedance control, enhancing robustness and stability. First, the excavator’s coupled nonlinear dynamics are transformed into an open-loop port Hamiltonian model with disturbances. Through an energy shaping method, this model becomes an ideal closed-loop port Hamiltonian system, stabilized asymptotically by damping injection. An improved robust disturbance observer estimates system disturbances, guiding control compensation term design. Hydraulic cylinder forces and displacements are calculated from the closed-loop port Hamiltonian system’s matching equations. By integrating passivity and impedance control, a flow controller resolves electrohydraulic servo system nonlinearity. Comparative analysis with existing methodologies demonstrates the proposed robust controller’s superior tracking accuracy, even in the presence of shock disturbances.
{"title":"Disturbance observer-based passivity and impedance control for trajectory tracking in autonomous hydraulic excavators","authors":"Junjie Gong, Jian Chen, Dengsheng Cai, Wei Wei, Yu Long","doi":"10.1016/j.autcon.2024.105898","DOIUrl":"https://doi.org/10.1016/j.autcon.2024.105898","url":null,"abstract":"Trajectory tracking control is pivotal for achieving autonomous operation in hydraulic excavators. This paper proposes a robust control scheme, merging passivity-based and impedance control, enhancing robustness and stability. First, the excavator’s coupled nonlinear dynamics are transformed into an open-loop port Hamiltonian model with disturbances. Through an energy shaping method, this model becomes an ideal closed-loop port Hamiltonian system, stabilized asymptotically by damping injection. An improved robust disturbance observer estimates system disturbances, guiding control compensation term design. Hydraulic cylinder forces and displacements are calculated from the closed-loop port Hamiltonian system’s matching equations. By integrating passivity and impedance control, a flow controller resolves electrohydraulic servo system nonlinearity. Comparative analysis with existing methodologies demonstrates the proposed robust controller’s superior tracking accuracy, even in the presence of shock disturbances.","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"52 1","pages":""},"PeriodicalIF":10.3,"publicationDate":"2024-12-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142867684","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-12-13DOI: 10.1016/j.autcon.2024.105926
Miyoung Uhm, Jaehee Kim, Seungjun Ahn, Hoyoung Jeong, Hongjo Kim
While Generative Pre-Trained Transformers (GPT)-based models offer high potential for context-specific information generation, inaccurate numerical responses, a lack of detailed information, and hallucination problems remain as the main challenges for their use in assisting safety engineering and management tasks. To address the challenges, this paper systematically evaluates the effectiveness of the Retrieval-Augmented Generation-based GPT (RAG-GPT) model for generating detailed and specific construction safety information. The RAG-GPT model was compared with four other GPT models, evaluating the models' responses from three different groups––2 researchers, 10 construction safety experts, and 30 construction workers. Quantitative analysis demonstrated that the RAG-GPT model showed superior performance compared to the other models. Experts rated the RAG-GPT model as providing more contextually relevant answers, with high marks for accuracy and essential information inclusion. The findings indicate that the RAG strategy, which uses vector data to enhance information retrieval, significantly improves the accuracy of construction safety information.
{"title":"Effectiveness of retrieval augmented generation-based large language models for generating construction safety information","authors":"Miyoung Uhm, Jaehee Kim, Seungjun Ahn, Hoyoung Jeong, Hongjo Kim","doi":"10.1016/j.autcon.2024.105926","DOIUrl":"https://doi.org/10.1016/j.autcon.2024.105926","url":null,"abstract":"While Generative Pre-Trained Transformers (GPT)-based models offer high potential for context-specific information generation, inaccurate numerical responses, a lack of detailed information, and hallucination problems remain as the main challenges for their use in assisting safety engineering and management tasks. To address the challenges, this paper systematically evaluates the effectiveness of the Retrieval-Augmented Generation-based GPT (RAG-GPT) model for generating detailed and specific construction safety information. The RAG-GPT model was compared with four other GPT models, evaluating the models' responses from three different groups––2 researchers, 10 construction safety experts, and 30 construction workers. Quantitative analysis demonstrated that the RAG-GPT model showed superior performance compared to the other models. Experts rated the RAG-GPT model as providing more contextually relevant answers, with high marks for accuracy and essential information inclusion. The findings indicate that the RAG strategy, which uses vector data to enhance information retrieval, significantly improves the accuracy of construction safety information.","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"29 1","pages":""},"PeriodicalIF":10.3,"publicationDate":"2024-12-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142816505","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-12-12DOI: 10.1016/j.autcon.2024.105907
Yixuan Chen, Sicong Xie, Jian Zhang
The performance of existing structures is often compromised by damage and condition changes, challenging current evaluation methods in accurately assessing their service status. This paper introduces a structural performance evaluation method via digital-physical twin and multi-parameter identification. Key features include: (1) a digital twin framework that integrates non-contact sensing data with finite element models. (2) a technique for local stiffness reduction using intelligent crack inspection data, where deep learning extracts crack information and a mechanical model calculates stiffness reduction coefficients. (3) a multi-parameter identification approach combining non-contact monitoring data with twin substructure models, employing substructure interaction technology and an enhanced unscented Kalman filter algorithm to identify critical parameters like support stiffness. The method's feasibility is demonstrated through a case study involving a frame structure, offering a new paradigm for the safety assessment of existing structures.
{"title":"Structural performance evaluation via digital-physical twin and multi-parameter identification","authors":"Yixuan Chen, Sicong Xie, Jian Zhang","doi":"10.1016/j.autcon.2024.105907","DOIUrl":"https://doi.org/10.1016/j.autcon.2024.105907","url":null,"abstract":"The performance of existing structures is often compromised by damage and condition changes, challenging current evaluation methods in accurately assessing their service status. This paper introduces a structural performance evaluation method via digital-physical twin and multi-parameter identification. Key features include: (1) a digital twin framework that integrates non-contact sensing data with finite element models. (2) a technique for local stiffness reduction using intelligent crack inspection data, where deep learning extracts crack information and a mechanical model calculates stiffness reduction coefficients. (3) a multi-parameter identification approach combining non-contact monitoring data with twin substructure models, employing substructure interaction technology and an enhanced unscented Kalman filter algorithm to identify critical parameters like support stiffness. The method's feasibility is demonstrated through a case study involving a frame structure, offering a new paradigm for the safety assessment of existing structures.","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"10 1","pages":""},"PeriodicalIF":10.3,"publicationDate":"2024-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142816507","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-12-12DOI: 10.1016/j.autcon.2024.105927
Antoine Gros, Livio De Luca, Frédéric Dubois, Philippe Véron, Kévin Jacquot
The assessment of structural safety and a thorough understanding of buildings' structural behavior are critical to enhancing the resilience of the built environment. Cultural Heritage (CH) buildings present unique diagnosis challenges due to their diverse designs and construction techniques, often requiring attention during maintenance or disaster relief efforts. However, collaboration across CH and Architecture, Engineering, and Construction (AEC) fields is hindered by increasing information complexity and prolonged feedback loops. This paper introduces a methodological approach utilizing Knowledge Graph technologies to integrate structural diagnosis information and processes. The approach is applied to the diagnosis of the Notre-Dame de Paris buttressing system, demonstrated through a proof-of-concept knowledge system. By leveraging Knowledge Graph functionalities, insights are derived from the spatialization and provenance of mechanical phenomena, including observed or simulation-predicted cracks in mortar-bound masonry.
{"title":"From surveys to simulations: Integrating Notre-Dame de Paris' buttressing system diagnosis with knowledge graphs","authors":"Antoine Gros, Livio De Luca, Frédéric Dubois, Philippe Véron, Kévin Jacquot","doi":"10.1016/j.autcon.2024.105927","DOIUrl":"https://doi.org/10.1016/j.autcon.2024.105927","url":null,"abstract":"The assessment of structural safety and a thorough understanding of buildings' structural behavior are critical to enhancing the resilience of the built environment. Cultural Heritage (CH) buildings present unique diagnosis challenges due to their diverse designs and construction techniques, often requiring attention during maintenance or disaster relief efforts. However, collaboration across CH and Architecture, Engineering, and Construction (AEC) fields is hindered by increasing information complexity and prolonged feedback loops. This paper introduces a methodological approach utilizing Knowledge Graph technologies to integrate structural diagnosis information and processes. The approach is applied to the diagnosis of the Notre-Dame de Paris buttressing system, demonstrated through a proof-of-concept knowledge system. By leveraging Knowledge Graph functionalities, insights are derived from the spatialization and provenance of mechanical phenomena, including observed or simulation-predicted cracks in mortar-bound masonry.","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"86 1","pages":""},"PeriodicalIF":10.3,"publicationDate":"2024-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142816508","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-12-12DOI: 10.1016/j.autcon.2024.105905
Boyi Duan, Kun Qian, Aohua Liu, Shan Luo
Cable-in-duct installation is one of the most challenging contact-rich interior finishing tasks for construction robots. Such precise robotic cable manipulation skills are expected to be endowed with high adaptability towards unstructured on-site construction activities via Sim2Real transfer. This paper presents a Sim2Real transferable reinforcement learning (RL) policy learning method for multi-stage robotic cable-in-duct installation, employing reward shaping to support unified task completion through a multi-stage RL policy. Specifically, the Foreground-aware Siamese Tactile Regression Network (FSTR-Net) is introduced as a feature-level unsupervised domain adaptation method to enhance the Sim2Real transfer of the RL strategy. Evaluations demonstrate that the robotic skill for cable-in-duct installation attains a success rate exceeding 98% in the simulator. FSTR-Net achieves over 99% accuracy for tactile-based in-hand fish tape pose estimation. Furthermore, real-world experiments show an average success rate of 95.8%, validating the RL strategy’s generalization and the approach’s effectiveness in mitigating the domain gap.
{"title":"Visual–tactile learning of robotic cable-in-duct installation skills","authors":"Boyi Duan, Kun Qian, Aohua Liu, Shan Luo","doi":"10.1016/j.autcon.2024.105905","DOIUrl":"https://doi.org/10.1016/j.autcon.2024.105905","url":null,"abstract":"Cable-in-duct installation is one of the most challenging contact-rich interior finishing tasks for construction robots. Such precise robotic cable manipulation skills are expected to be endowed with high adaptability towards unstructured on-site construction activities via Sim2Real transfer. This paper presents a Sim2Real transferable reinforcement learning (RL) policy learning method for multi-stage robotic cable-in-duct installation, employing reward shaping to support unified task completion through a multi-stage RL policy. Specifically, the Foreground-aware Siamese Tactile Regression Network (FSTR-Net) is introduced as a feature-level unsupervised domain adaptation method to enhance the Sim2Real transfer of the RL strategy. Evaluations demonstrate that the robotic skill for cable-in-duct installation attains a success rate exceeding 98% in the simulator. FSTR-Net achieves over 99% accuracy for tactile-based in-hand fish tape pose estimation. Furthermore, real-world experiments show an average success rate of 95.8%, validating the RL strategy’s generalization and the approach’s effectiveness in mitigating the domain gap.","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"8 1","pages":""},"PeriodicalIF":10.3,"publicationDate":"2024-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142816506","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}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：