Abstract
Abstract. Construction site monitoring is an essential task for keeping track of the ongoing construction work and providing up-to-date information for a Building Information Model (BIM). The BIM contains the as-planned states (geometry, schedule, costs, ...) of a construction project. For updating, the as-built state has to be acquired repeatedly and compared to the as-planned state. In the approach presented here, a 3D representation of the as-built state is calculated from photogrammetric images using multi-view stereo reconstruction. On construction sites one has to cope with several difficulties like security aspects, limited accessibility, occlusions or construction activity. Different acquisition strategies and techniques, namely (i) terrestrial acquisition with a hand-held camera, (ii) aerial acquisition using a Unmanned Aerial Vehicle (UAV) and (iii) acquisition using a fixed stereo camera pair at the boom of the crane, are tested on three test sites. They are assessed considering the special needs for the monitoring tasks and limitations on construction sites. The three scenarios are evaluated based on the ability of automation, the required effort for acquisition, the necessary equipment and its maintaining, disturbance of the construction works, and on the accuracy and completeness of the resulting point clouds. Based on the experiences during the test cases the following conclusions can be drawn: Terrestrial acquisition has the lowest requirements on the device setup but lacks on automation and coverage. The crane camera shows the lowest flexibility but the highest grade of automation. The UAV approach can provide the best coverage by combining nadir and oblique views, but can be limited by obstacles and security aspects. The accuracy of the point clouds is evaluated based on plane fitting of selected building parts. The RMS errors of the fitted parts range from 1 to a few cm for the UAV and the hand-held scenario. First results show that the crane camera approach has the potential to reach the same accuracy level.
Highlights
1.1 MotivationConstruction progress monitoring is the comparison of planned states of the construction to the actual state at a certain time step
Detected deviations from the planned states are intended to lead to modifications of the schedule and the following processes modeled in the Building Information Models (BIM)
There are several additional works related to the approach of (GolparvarFard et al, 2012) (e.g., (Karsch et al, 2014, Ham et al, 2016)), but to the knowledge of the authors this is the only group working on construction site monitoring using photogrammetric point clouds
Summary
Construction progress monitoring is the comparison of planned states (as-planned) of the construction to the actual state (as-built) at a certain time step. Today this is a primarily manual process which is usually based on 2D plans. A BIM is a digital representation of a built or planned building that stores its 3D geometry, and embeds the temporal information such as the scheduling of the 4D construction process. Remote sensing techniques (i.e., laser scanning or image-based techniques) are suited for the (automated) acquisition of the as-built state, since they do not need physical access to the building elements, like methods which require for example the mounting of RFID-markers. The flexibility of image-based techniques is the motivation to investigate these alternatives to approaches using terrestrial laser scanning (TLS)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
