Haili Jiang
Haili Jiang has a total of 23 co-authors. Here are the top co-authors by number of shared publications:
- Dejiang Wang
Department of Civil Engineering, School of Mechanics and Engineering Science, Shanghai University, Shanghai 200444, China
4 shared publications - Panpan Liu
Shanghai Highway and Bridge (Group) Co., Ltd., Shanghai 200433, China
4 shared publications - Bolong Liu
Department of Civil Engineering, School of Mechanics and Engineering Science, Shanghai University, Shanghai 200444, China
1 shared publication - Dong Xu
College of Civil Engineering, Tongji University, Shanghai 200092, China
1 shared publication - Haoyu Wang
Department of Civil Engineering, School of Mechanics and Engineering Science, Shanghai University, Shanghai 200444, China
1 shared publication Huili Wang
1 shared publication
Latest Publications
- Open Access
Automated Recognition and Measurement of Corrugated Pipes for Precast Box Girder Based on RGB-D Camera and Deep Learning
The accurate installation position of corrugated pipes is critical for ensuring the quality of prestressed concrete box girders. Given that these pipes can span up to 30 m and are deeply embedded within rebars, manual measurement is both labor-intensive and prone to errors. Meanwhile, automated recognition and measurement methods are hindered by high equipment costs and accuracy issues caused by rebar occlusion. To balance cost effectiveness and measurement accuracy, this paper proposes a method that utilizes an RGB-D camera and deep learning. Firstly, an optimal registration scheme is selected to generate complete point cloud data of pipes from segmented data captured by an RGB-D camera. Next, semantic segmentation is applied to extract the characteristic features of the pipes. Finally, the center points from cross-sectional slices are extracted and curve-fitting is performed to recognize and measure the pipes. A test was conducted in a simulated precast factory environment to validate the proposed method. The results show that under the optimal fitting scheme (BP neural network with circle fitting constraint), the average measurement errors for the three pipes are 2.2 mm, 1.4 mm, and 1.6 mm, with Maximum Errors of −5.8 mm, −4.2 mm, and −5.7 mm, respectively, meeting the standard requirements. The proposed method can accurately locate the pipes, offering a new technical pathway for the automated recognition and measurement of pipes in prefabricated construction.
Open AccessAutomated Recognition and Measurement of Corrugated Pipes for Precast Box Girder Based on RGB-D Camera and Deep Learning
The accurate installation position of corrugated pipes is critical for ensuring the quality of prestressed concrete box girders. Given that these pipes can span up to 30 m and are deeply embedded within rebars, manual measurement is both labor-intensive and prone to errors. Meanwhile, automated recognition and measurement methods are hindered by high equipment costs and accuracy issues caused by rebar occlusion. To balance cost effectiveness and measurement accuracy, this paper proposes a method that utilizes an RGB-D camera and deep learning. Firstly, an optimal registration scheme is selected to generate complete point cloud data of pipes from segmented data captured by an RGB-D camera. Next, semantic segmentation is applied to extract the characteristic features of the pipes. Finally, the center points from cross-sectional slices are extracted and curve-fitting is performed to recognize and measure the pipes. A test was conducted in a simulated precast factory environment to validate the proposed method. The results show that under the optimal fitting scheme (BP neural network with circle fitting constraint), the average measurement errors for the three pipes are 2.2 mm, 1.4 mm, and 1.6 mm, with Maximum Errors of −5.8 mm, −4.2 mm, and −5.7 mm, respectively, meeting the standard requirements. The proposed method can accurately locate the pipes, offering a new technical pathway for the automated recognition and measurement of pipes in prefabricated construction.
- Open Access
Path Planning for Construction Robot Based on the Improved A* Algorithm and Building Information Modeling
Path planning is a key technology in the automation of construction robots, and its implementation mainly relies on navigation algorithms and the generation of environment maps. However, the original A* algorithm has issues such as not considering dimensional constraints, a high number of turning nodes, and unsmooth paths. Moreover, traditional environment map generation is time-consuming and energy-consuming. Therefore, this paper proposes a path-planning method for construction robots based on an improved A* algorithm and BIM. First, the A* algorithm is improved by incorporating dimensional constraints for both the building and the robot and by reasonably setting virtual obstacles to enhance the feasibility of robot movement. Additionally, B-spline curves are used to optimize the path, reduce turning points, and achieve a smoother trajectory. Second, Dynamo visual programming is used to process the building’s BIM model, enabling efficient generation of the environment map while preserving the rich semantic and geometric information of the building. Finally, the improved A* algorithm is employed to perform path planning within the environment map generated from the BIM model, resulting in the generation of the optimal path. Through experimental verification on the civil engineering building of a university, the results show that the improved A* algorithm can successfully navigate within the BIM-generated map. It not only ensures the safety and smoothness of path planning but also enables the fast and efficient generation of environment maps. This effectively advances the research and application of BIM technology in the field of robot navigation.
Open AccessPath Planning for Construction Robot Based on the Improved A* Algorithm and Building Information Modeling
Path planning is a key technology in the automation of construction robots, and its implementation mainly relies on navigation algorithms and the generation of environment maps. However, the original A* algorithm has issues such as not considering dimensional constraints, a high number of turning nodes, and unsmooth paths. Moreover, traditional environment map generation is time-consuming and energy-consuming. Therefore, this paper proposes a path-planning method for construction robots based on an improved A* algorithm and BIM. First, the A* algorithm is improved by incorporating dimensional constraints for both the building and the robot and by reasonably setting virtual obstacles to enhance the feasibility of robot movement. Additionally, B-spline curves are used to optimize the path, reduce turning points, and achieve a smoother trajectory. Second, Dynamo visual programming is used to process the building’s BIM model, enabling efficient generation of the environment map while preserving the rich semantic and geometric information of the building. Finally, the improved A* algorithm is employed to perform path planning within the environment map generated from the BIM model, resulting in the generation of the optimal path. Through experimental verification on the civil engineering building of a university, the results show that the improved A* algorithm can successfully navigate within the BIM-generated map. It not only ensures the safety and smoothness of path planning but also enables the fast and efficient generation of environment maps. This effectively advances the research and application of BIM technology in the field of robot navigation.
- Open Access
Existing Buildings Recognition and BIM Generation Based on Multi-Plane Segmentation and Deep Learning
Point cloud-based BIM reconstruction is an effective approach to enabling the digital documentation of existing buildings. However, current methods often demand substantial time and expertise for the manual measurement of building dimensions and the drafting of BIMs. This paper proposes an automated approach to BIM modeling of the external surfaces of existing buildings, aiming to streamline the labor-intensive and time-consuming processes of manual measurement and drafting. Initially, multi-angle images of the building are captured using drones, and the building’s point cloud is reconstructed using 3D reconstruction software. Next, a multi-plane segmentation technique based on the RANSAC algorithm is applied, facilitating the efficient extraction of key features of exterior walls and planar roofs. The orthophotos of the building façades are generated by projecting wall point clouds onto a 2D plane. A lightweight convolutional encoder–decoder model is utilized for the semantic segmentation of windows and doors on the façade, enabling the precise extraction of window and door features and the automated generation of AutoCAD elevation drawings. Finally, the extracted features and segmented data are integrated to generate the BIM. The case study results demonstrate that the proposed method exhibits a stable error distribution, with model accuracy exceeding architectural industry requirements, successfully achieving reliable BIM reconstruction. However, this method currently faces limitations in dealing with buildings with complex curved walls and irregular roof structures or dense vegetation obstacles.
Open AccessExisting Buildings Recognition and BIM Generation Based on Multi-Plane Segmentation and Deep Learning
Point cloud-based BIM reconstruction is an effective approach to enabling the digital documentation of existing buildings. However, current methods often demand substantial time and expertise for the manual measurement of building dimensions and the drafting of BIMs. This paper proposes an automated approach to BIM modeling of the external surfaces of existing buildings, aiming to streamline the labor-intensive and time-consuming processes of manual measurement and drafting. Initially, multi-angle images of the building are captured using drones, and the building’s point cloud is reconstructed using 3D reconstruction software. Next, a multi-plane segmentation technique based on the RANSAC algorithm is applied, facilitating the efficient extraction of key features of exterior walls and planar roofs. The orthophotos of the building façades are generated by projecting wall point clouds onto a 2D plane. A lightweight convolutional encoder–decoder model is utilized for the semantic segmentation of windows and doors on the façade, enabling the precise extraction of window and door features and the automated generation of AutoCAD elevation drawings. Finally, the extracted features and segmented data are integrated to generate the BIM. The case study results demonstrate that the proposed method exhibits a stable error distribution, with model accuracy exceeding architectural industry requirements, successfully achieving reliable BIM reconstruction. However, this method currently faces limitations in dealing with buildings with complex curved walls and irregular roof structures or dense vegetation obstacles.
- 26 November 2023
- Published by Elsevier in Ocean Engineering
- Vol. 291
Anti-scour performance of fluidized solidified slurry in dynamic water for scour repair
- 26 November 2023
- Published by Elsevier in Ocean Engineering
- Vol. 291
Anti-scour performance of fluidized solidified slurry in dynamic water for scour repair
- Open Access
- 12 May 2023
- Published by MDPI AG in Sustainability
- Vol. 15 (10) , 7951
Sustainable Operations: A Systematic Operational Performance Evaluation Framework for Public–Private Partnership Transportation Infrastructure Projects
With the application of public–private partnership (PPP) model in urban transportation infrastructure projects, various participants have put forward multi-dimensional demands to the operation and maintenance of infrastructures. This study aims to establish a systematic operational performance evaluation framework for PPP transportation infrastructure projects. Based on a literature review, the balanced scorecard was improved, and a conceptual model of multidimensional performance assessment was constructed. The structure of the qualitative performance assessment system was quantified and analyzed by combining structural equation modeling with questionnaires to obtain causal relationships among the indicators. Subsequently, a system dynamics model was constructed to assess the performance dynamically, and a validation analysis was conducted. It finds that maintaining a low level of operational quality over an extended period can significantly reduce stakeholder satisfaction, consequently exacerbating the decline in project performance. In contrast, an improvement in the level of informatization is found to positively contribute to enhancing operational quality and facilitating the long-term sustainability of project operations. It innovatively integrates four dimensions of financial, multi-stakeholder satisfaction, operation and maintenance quality, and sustainability performance to enrich the theoretical system of PPP transportation infrastructure performance assessment. At the same time, it analyzes the influence mechanism among the indicators and its long-term dynamic performance, which provides an effective decision-making tool for operational performance management.
- 12 May 2023
- Published by MDPI AG in Sustainability
- Vol. 15 (10) , 7951
Open AccessSustainable Operations: A Systematic Operational Performance Evaluation Framework for Public–Private Partnership Transportation Infrastructure Projects
With the application of public–private partnership (PPP) model in urban transportation infrastructure projects, various participants have put forward multi-dimensional demands to the operation and maintenance of infrastructures. This study aims to establish a systematic operational performance evaluation framework for PPP transportation infrastructure projects. Based on a literature review, the balanced scorecard was improved, and a conceptual model of multidimensional performance assessment was constructed. The structure of the qualitative performance assessment system was quantified and analyzed by combining structural equation modeling with questionnaires to obtain causal relationships among the indicators. Subsequently, a system dynamics model was constructed to assess the performance dynamically, and a validation analysis was conducted. It finds that maintaining a low level of operational quality over an extended period can significantly reduce stakeholder satisfaction, consequently exacerbating the decline in project performance. In contrast, an improvement in the level of informatization is found to positively contribute to enhancing operational quality and facilitating the long-term sustainability of project operations. It innovatively integrates four dimensions of financial, multi-stakeholder satisfaction, operation and maintenance quality, and sustainability performance to enrich the theoretical system of PPP transportation infrastructure performance assessment. At the same time, it analyzes the influence mechanism among the indicators and its long-term dynamic performance, which provides an effective decision-making tool for operational performance management.
- Open Access
Synthetic Datasets for Rebar Instance Segmentation Using Mask R-CNN
The construction and inspection of reinforcement rebar currently rely entirely on manual work, which leads to problems such as high labor requirements and labor costs. Rebar image detection using deep learning algorithms can be employed in construction quality inspection and intelligent construction; it can check the number, spacing, and diameter of rebar on a construction site, and guide robots to complete rebar tying. However, the application of deep learning algorithms relies on a large number of datasets to train models, while manual data collection and annotation are time-consuming and laborious. In contrast, using synthetic datasets can achieve a high degree of automation of annotation. In this study, using rebar as an example, we proposed a mask annotation methodology based on BIM software and rendering software, which can establish a large and diverse training set for instance segmentation, without manual labeling. The Mask R-CNN trained using both real and synthetic datasets demonstrated a better performance than the models trained using only real datasets or synthetic datasets. This synthetic dataset generation method could be widely used for various image segmentation tasks and provides a reference for other computer vision engineering tasks and deep learning tasks in related fields.
Open AccessSynthetic Datasets for Rebar Instance Segmentation Using Mask R-CNN
The construction and inspection of reinforcement rebar currently rely entirely on manual work, which leads to problems such as high labor requirements and labor costs. Rebar image detection using deep learning algorithms can be employed in construction quality inspection and intelligent construction; it can check the number, spacing, and diameter of rebar on a construction site, and guide robots to complete rebar tying. However, the application of deep learning algorithms relies on a large number of datasets to train models, while manual data collection and annotation are time-consuming and laborious. In contrast, using synthetic datasets can achieve a high degree of automation of annotation. In this study, using rebar as an example, we proposed a mask annotation methodology based on BIM software and rendering software, which can establish a large and diverse training set for instance segmentation, without manual labeling. The Mask R-CNN trained using both real and synthetic datasets demonstrated a better performance than the models trained using only real datasets or synthetic datasets. This synthetic dataset generation method could be widely used for various image segmentation tasks and provides a reference for other computer vision engineering tasks and deep learning tasks in related fields.
- 1 January 2023
- Published by International Association for Bridge and Structural Engineering (IABSE)
- Vol. 119, 1003-1009
Stress Indicators - A Complete Checking System of Stresses for Box Section
The Box section has complicated mechanical behavior under loads, including in-plane and out-of-plane stresses in each slab of box components. Traditional checking stresses originated from simple open sections, such as rectangular, T and I sections, featured by normal stresses at extreme fibers and shear stress in the web. For statically indeterminate box sections, cracks found in the outside surface of the web were much different from those found inside the web. This means that the stress indicator for the outside surface of the web should be different from the stress indicator for the inside surface of the web. These indicators are even different from the indicator at the middle lane of the web, which actually is the traditional shear stress indication in the web. The newly developed stress indicator system includes nine stress indicators in each slab of the box components. Each slab has three layers, i.e., the top, bottom and middle layer. Each layer has three stress directions: longitudinal, transverse and principal. Thus, the nine stress indicators for each slab will consist of all the mechanical behavior, including bending, shear and torsion, under symmetrical and unsymmetrical loads for slabs of the box section. This stress indicator system plays an important role in forensic engineering, especially for deficient bridges.
- 1 January 2023
- Published by International Association for Bridge and Structural Engineering (IABSE)
- Vol. 119, 1003-1009
Stress Indicators - A Complete Checking System of Stresses for Box Section
The Box section has complicated mechanical behavior under loads, including in-plane and out-of-plane stresses in each slab of box components. Traditional checking stresses originated from simple open sections, such as rectangular, T and I sections, featured by normal stresses at extreme fibers and shear stress in the web. For statically indeterminate box sections, cracks found in the outside surface of the web were much different from those found inside the web. This means that the stress indicator for the outside surface of the web should be different from the stress indicator for the inside surface of the web. These indicators are even different from the indicator at the middle lane of the web, which actually is the traditional shear stress indication in the web. The newly developed stress indicator system includes nine stress indicators in each slab of the box components. Each slab has three layers, i.e., the top, bottom and middle layer. Each layer has three stress directions: longitudinal, transverse and principal. Thus, the nine stress indicators for each slab will consist of all the mechanical behavior, including bending, shear and torsion, under symmetrical and unsymmetrical loads for slabs of the box section. This stress indicator system plays an important role in forensic engineering, especially for deficient bridges.