Search

The challenge that this paper addresses is how to generate geometric digital twins of the indoor environment of buildings automatically. Unlike most previous research that starts with detecting planes in the point cloud and considers only geometric information, the proposed ‘void-growing’ approach is a full-automatic approach that starts with detecting void space inside rooms, considering geometric information, as well as semantic information predicted from deep learning. Then, based on the detected room spaces, structural elements, as well as doors and windows, are extracted. The method can work in (a) rooms with complex structures like U-shape and L-shape, (b) rooms with different ceiling heights and (c) rooms under a high occlusion level. Compared with previous studies that mainly use geometric information only, the approach also focuses on how to select useful information predicted by deep learning. This study used existing state-of-the-art deep learning architecture for the segmentation task in the proposed approach. By taking useful semantic information into consideration, the proposed approach performs better in creating geometric digital twins of buildings.

Scanning is a key element for many use cases in the Architectural, Engineering, Construction and Operation industry. It provides point clouds used for construction quality assurance, scan-to-BIM workflows and construction surveys. However, data acquisition using static laser scanners or photogrammetry methods is labour intensive during scanning and post-processing. Mobile scanners are conceptually the solution to this problem, given their potential to dramatically reduce on-site scanning effort and eliminate post-processing work. However, current mobile mapping devices are limited to producing point clouds of relatively low resolution. In this paper, we propose a dense 3D reconstruction pipeline for improving the resolution of point clouds, suitable for hand-held scanners comprised of a colour camera and a lidar. We fuse time-synchronized and spatially registered images and lidar sweeps using a deep learning method into dense scans of higher resolution, which are then used for progressive reconstruction. The novelty of our approach is that we first increase the precision and density of a bunch of individual lidar scans by inferring additional geometric constraints coming from predicted feature maps in the corresponding images. Then we automatically register these scans together, thus reconstructing the scene progressively in an odometric manner. We built a prototypic scanner, implemented our reconstruction pipeline as a software package and tested the whole in both indoor and outdoor case studies. The results showed that our method provided an overall noise reduction in point clouds by 11% and increased their density around 6 times.

While GPS and RFID are dominantly used for tracking construction equipment and materials, 3D vision-based tracking that employs stereo-vision system has been also proposed as an alternative to efficiently track construction resources operating in a congested site. The 3D vision-based tracking requires an entity to be tracked in both cameras of the stereovision system so that two of 2D pixel coordinates are generated for each entity. Its 3D coordinate can be calculated by triangulating two projection rays that go through the 2D pixel coordinates. In order to track multiple entities of the same type simultaneously, the triangulation requires an additional process to match entities across the two camera views. This paper proposes an efficient method of entity matching for tracking construction workers. The method is mainly based on epipolar geometry which represents the geometric relation between two camera views. Two additional strategies - distance ratio thresholding and frame grouping - are introduced to reject false matchings and to ultimately increase precision of the 3D tracking. Experiment results demonstrate its suitability for automated entity matching for 3D vision-based tracking of construction workers.

Accurate assessment of progress allows managers to make adjustments to minimise costs when deviations from the schedule occur. Current practices of predicting the performance of a construction team require inspections that are still mainly manual, time consuming and can contain errors. Improper understanding of the current status can lead to further errors and unsuitable adjustments by the managing team, leading to more delays and increased cost. The problem is amplified when inspections take place in interior environments where the tasks are more complex. Efficient progress monitoring systems can help automate progress inspections, reduce the risks of error, facilitate proper and timely corrective actions, and prevent deviations in terms of cost and schedule. This paper presents a literature synthesis and interpretation of the state-of-the-art progress monitoring methods in terms of (a) data acquisition, as in technologies that are used for capturing the as-built scenes, (b) information retrieval, as in extracting the information needed from the as-built data, (c) progress estimation, which includes the comparison between the as-built and as-planned model in order to define the progress status and (d) visualisation of the results. The methods reviewed were categorised in terms of the technology used and assessed in terms of utility, time efficiency, accuracy, level of automation, required preparation, training requirements, cost and mobility. The review concludes with a recommendation of the most appropriate technologies to use based on the type of activities, environment and the needs of the inspection.

The digitisation of existing railway geometry from point clouds referred to as “twinning” is a laborious task; currently outweighing the perceived benefits of the resulting model. State-of-the-art methods have provided promising results, yet they cannot offer large-scale rail class segmentation requires over kilo-metres without forfeiting precision and labour cost. The authors exploit the potential benefits of railway topology to automate the twinning process. The preliminary step is automatically segmenting mast point clusters as their positions are critical for the subsequent railway assets’ class segmentation. The proposed method first removes vegetation and noise; then segments masts using the RANSAC algorithm relative to the track centerline, and delivers final models in IFC format. The authors validated the method on 18 km railway point cloud and yielded an overall segmentation accuracy of 90.1% F1 score. The proposed method lays foundations to efficiently generate geometry-only digital twins of railway assets with no prior information.

The automated generation of geometry-only digital twins of Overhead Line Equipment (OLE) system in existing railways from point clouds is an unsolved problem. Currently, this process is highly reliant upon manual inputs, needing 10 times more labour hours than scanning the physical asset. The resulting modelling cost counteracts the expected benefits of the digital twin. We tackle this challenge using a novel model-driven method that exploits the highly regulated and standardised nature of railways. It starts by restricting the search for OLE elements relative to point clusters of the railway masts. The resulting point clusters of the OLE elements are then converged with various parametric models of different catenary configurations to verify the presence of OLE elements and to find the best possible fit. The method outputs a geometry-only digital twin of the OLE system in Industry Foundation Classes (IFC) format. The method was tested on an 18 km railway point cloud and achieves overall detection rates of 93.2% F1 score for OLE cables and 98.1% F1 score for other OLE elements. The accuracy of the generated model is evaluated using distance-based metrics between the ground truth model and the automated model. The average modelling distance is 3.82 cm Root Mean Square Error (RMSE) for all 18 km data

© 2016 Elsevier B.V. Tracking construction workers onsite yields trajectory data that can be very useful in safety and productivity monitoring. Tracking via cameras and computer vision-based methods is one of the tracking options available and deemed suitable for worker tracking in congested outdoor environments. However, this option has yet to yield a practical, robust solution, because of its poor performance in handling occlusions and maintaining trajectories over time. This is due, on one hand, to the tracking methods' degeneracy issues, and the use of detection methods only for initializing trackers. On the other hand, detection methods alone are unable to yield robust trajectories because they treat each frame independently and do not account for temporal effects. Furthermore, no detection method satisfies high performance requirements in both precision and recall. Higher precision entails lower recall. False positives or false negatives limit practical applications. To resolve these issues, this paper presents a hybrid method for locating construction workers that fuses tracking and detection together throughout the tracking process. The detection method is designed to hit higher precision while the tracking method supplements the reduced lower recall. Tracking allows entity matching across frames producing trajectories for all identified workers, while tracking of an occluded worker is automatically terminated and resumed when the occlusion gets cleared. Experimental tests on onsite videos demonstrate that the proposed combination improves overall tracking performance.

Report of the International Workshop on Built Environment Digital Twinning presented by TUM Institute for Advanced Study and Siemens AG

The design development, manufacture and construction process for a new bridge can be a lengthy and iterative process. The current design process through construction is virtually unmapped as the majority of structures in the built environment are unique. In many cases, reasonable assumptions and changes need to be made in the absence of the appropriate information and decisions at the appropriate time. However, the assumptions and changes made during the design process rely on the experience of engineers, making the process difficult to be analyzed and improved. There is therefore a need for a new approach to enable explicit mapping of the decision making process to facilitate efficient optimization through design. In order to address this problem, this study presents a novel approach to streamline the design decision making process of bridges. The key theme of this approach is to map the bridge design decision making processes and analyze them using digital engineering tools. Design Structure Matrix (DSM) is used in this study as a digital tool to map and optimize the current bridge design decision making process. As the means of validating the proposed approach, a real-life bridge project, which is a typical highway bridge of the UK, is identified for investigation. The result of the case study based on the proposed approach shows a promising potential that the current bridge design process can be potentially improved in an explicit and efficient way and the captured stages may be simulated through design automation tools, Innovate UK part-funding

The Precedence Diagramming Method (PDM) is the prevalent scheduling technique used for temporal planning of projects. Critical paths require special attention during the course of planning as their lengths define the project duration. However, increasing/decreasing the duration of an activity on the critical path does not necessarily results in a longer/shorter project duration. Responses of the critical path in a PDM network for small changes of the duration of a critical activity can be classified into six categories, while nine different possible categories exist in theory. The modeling capabilities of PDM recently have been extended by the introduction of a) point-to-point relations, b) continuous relations, and c) non-linear activity (production-time) functions. The paper presents the following conjecture that needs a proof: The number of existing types of critical activities remains six when using the recently developed precedence relations and non-linear production-time functions for activities. The theory of the different types of critical activities is briefly discussed and the mathematical model of the generalized PDM technique is presented. An additional feature presented in the paper beyond defining the conjecture is the introduction of influence lines for project duration that have never been used in project management before. Influence lines can be of great help in understanding the nature of critical activities. Papers providing solutions or confuting this conjecture will be honored with the CCC 2017 award.

Collision of over-height vehicles with low bridges and tunnels occur with high frequency in the UK as many structures were built at a time when there was less moving traffic on the roadway. These older bridges are now considered at risk of vehicular strikes due to its low clearance height (less than 16 feet 6 inches or 5.03 metres). While previous methods have used vision-based systems to address the over-height warning problem, such methods are sensitive to wind. In this paper, we proposed an extension of the work done to minimise false detections due to wind by using a constraint-based method to track motion trajectories to improve the overall performance of the system. The dataset consists of 102 over-height vehicles recorded at 25 fps. The paper compares feature detectors to optimally track vehicle trajectories and analyses its motion to accurately classify positive detections. The final validation yields a performance of 94.5% recall and 91.1% precision., Career Integration Grants (CIG) - Marie Curie Actions

In addition to the traditional benefits that are associated with the installation of structural health monitoring systems, reductions in construction, operational, and maintenance costs and improved performance and quality can be achieved by effectively using the acquired data. However, considered in isolation, the raw data is of little use and value. It needs to be processed and put into a geometric context within the infrastructure asset, which facilitates the interpretation and analysis of the data. This supports informed decision making that leads to effective actions. This study outlines a new approach that enables to model structural performance monitoring systems in a BIM environment and hence permits visualising sensor data directly on BIM models. The paper addresses aspects related to (i) interoperability and standard data models, (ii) management and visualisation of monitoring data, and (iii) data interpretation and analysis. A prestressed concrete bridge, with a comprehensive built-in structural performance monitoring system, has been used as a case study. The case study demonstrates that by including and visualising monitoring data directly on BIM models, the acquired data gains geometrical context within the built asset, which facilitates better interpretation, analysis, and all the data-sharing benefits associated with the BIM approach.

Significant difficulties remain in exchanging information between building information modeling (BIM) tools. The industry foundation classes (IFC) exchange schema is too generic to capture the full semantic meaning needed for direct use by different construction project stakeholders’ BIM tools. Although BIM standards that prescribe model view definitions (MVD) for domain‐specific exchanges are under development, insufficient semantic definition of exchanges prevents achievement of the full potential of BIM through seamless interoperability. We propose an innovative approach for supplementing an IFC exchange file with semantically useful concepts inferred from the explicit and implicit information contained in the building model. A prototype software was implemented to test the applicability of the approach. It consists of a rule‐processing engine and allows composition of inference rule‐sets that can be tailored for different domains. The tests demonstrate semantic enrichment with precast concrete building models, adding inferred joint, slab aggregation and connection concepts