Derivation of Minimum Required Model for Augmented Reality Based Stepwise Construction Assembly Control
MetadataShow full item record
The global 3D imaging market is expected to reach $26 billion by 2024 with an annual growth of 23.7% (3D Imaging Market Global Scenario, Market Size, Trend and Forecast, 2015 – 2024. 2018). Various industries are extensively involved in this emergence including the healthcare and entertainment industries, the architecture and construction industries. Additionally, global steel pipe demand is predicted to rise by 3.5% annually until 2019. The combination of the two growths raises the potential of 3D imaging technologies in the construction industry, especially in the piping industry. Thanks to the virtuous cycle between growth and innovation, development and applications of new 3D vision technologies and techniques has become a need for the construction industry facing harsh competition globally. Similarly, prefabrication has been boosted in the construction industry, reducing costs and optimizing time of fabrication. It also copes with the increased demand of small tolerances which sets the industry and its labor under high pressure. Thus, quality control is reinforced in fabrication facilities, and innovations can be deployed in that domain to preclude assemblies from any incompliance. Employing 3D scanners is one effective way to do so, and the recent emergence of handheld laser scanners has created the opportunity to develop efficient new methods to be used for quality control. This thesis proposes a novel methodology for deriving 3D models for assemblies to be fabricated, breaking down a barrier that previously inhibited the utilization of small-range handheld 3D laser scanners. The process is applicable for industrial assembly lines, which present a stepwise fabrication process such as that for pipe spools. The methodology also aims at streamlining the fabrication flow for workers, and can provide as-built information to the management team. To do so, piping assemblies are thoroughly analyzed and decomposed at each and every step around the weld of interest: one part is being added with respect to the other. From this decomposition of a pipe spool, the challenge of the methodology is to shrink down to the minimum the amount of components that have to be investigated to control the geometry of the assembly. The key concept of solid of revolution is introduced and permits the derivation of the Minimum Required Model (MRM). Examples are generated and experiments are conducted to test the effectiveness of the presented method. This is mainly realized by implementing the algorithm within an in-house software, developed along with another PhD student, a master’s student and a co-op student. The software enables the comparison of the acquired scene to the 3D model by segmenting piping components individually, and generating the as-modelled point cloud. Consequently, piping components can directly be segmented within the software, and the MRM can be derived and compared to the expected model. In order to evaluate the efficiency of the method, three criteria are proposed: (1) the level of spatial complexity between the derived Minimum Required Model and the initial 3D model, (2) the capacity to use a handheld scanner with or without the MRM, and finally (3) the accuracy of the comparison between the acquired scan and the 3D model.
Cite this version of the work
Nicolas Jeanclos (2018). Derivation of Minimum Required Model for Augmented Reality Based Stepwise Construction Assembly Control. UWSpace. http://hdl.handle.net/10012/13461