The delay of construction projects and the resulting cost increase is a common issue in civil engineering projects. These difficulties are often related to poor site supervision and project performance management [66]. The advances in computer science and 3D geometry scanning allow the construction of a dense point cloud which can be used to automatically identify the presence and absence of objects on the construction site. The automation of construction site progress monitoring is still subject to research and includes time-consuming computations. These problems include the alignment of two data sets in a common coordinate system and object recognition which both offer great acceleration potential through parallel data processing [8]. The automatic progress detection pipeline is analysed and an approach for the application of graphics processing units (GPU) to accelerate the processes of rigid point cloud alignment and building component classification is proposed. Implementations based on NVIDIA Cuda originally proposed by Tamaki et al. [58] for point cloud alignment and a new method for progress detection implemented with Microsoft's C++ AMP are compared with current automated progress tracking procedures. The evaluations are performed using data provided by Braun et al. [13] which was acquired on two different construction sites in Munich.     «

The delay of construction projects and the resulting cost increase is a common issue in civil engineering projects. These difficulties are often related to poor site supervision and project performance management [66]. The advances in computer science and 3D geometry scanning allow the construction of a dense point cloud which can be used to automatically identify the presence and absence of objects on the construction site. The automation of construction site progress monitoring is still subjec...     »