Graduate Thesis Or Dissertation
 

Automated Monitoring of Modular Construction Factories using Computer Vision

Public Deposited

Downloadable Content

Download PDF
https://ir.library.oregonstate.edu/concern/graduate_thesis_or_dissertations/7p88cq99f

Descriptions

Attribute NameValues
Creator
Abstract
  • Modular construction is increasingly seen as an efficient construction method in terms of time, cost, and energy. The full realization of these advantages partly relies on the efficiency of the production process inside the modular factories, which currently rely on tedious manual monitoring methods or expensive automated techniques. As a result, major bottlenecks are commonly formed on the production line, which can significantly delay construction projects. Computer vision-based methods have recently emerged as suitable methods to monitor modular factories by focusing on the progress tracking of panelized units and ergonomic posture assessment. However, there is a gap in knowledge regarding the translation of video data to operational insights that can inform resource allocation and bottleneck prediction in the assembly line. This research aims to fill this gap by proposing computer vision-based methodologies to monitor resource usage and proactively predict productivity bottlenecks in modular construction factories. To achieve this goal, this study proposes methodologies to: (1) detect the presence of workers and identify their activity using human activity recognition methods in the context of modular construction factories; (2) monitor the installation of subassemblies inside modular construction factories by integrating Building Information Modeling with computer vision and computer graphics techniques; and (3) identify bottlenecks by proposing a computer vision-based data-driven bottleneck detection methodology. The proposed methodologies have been validated using videos captured from a modular construction factory in the U.S. demonstrating their effectiveness and applicability. This research demonstrates the feasibility of computer vision-based methods to automatically monitor key resources inside modular construction factories and aid productivity management. This study contributes to the body of knowledge by providing means for automated data acquisition and monitoring of modular factories using computer vision. The study also makes practical contributions by validating the proposed methodologies in real-world scenarios and identifying the relevant challenges. Specifically, the main contributions of this research are: (1) identifying the challenges that modular factories pose for computer vision-based worker monitoring methods; (2) addressing the technical challenges posed by occlusions inside modular factories; and (3) proposing a novel computer vision-based data-driven bottleneck detection method in modular construction factories. The proposed methodologies in this research can be expanded to comprehensively monitor the workers’ activities and the installation of subassemblies and can be used to identify bottlenecks in different factories and assist the plant manager in decision-making.
  • Keywords: Activity Recognition, Bottleneck, Offsite Construction, Ray casting, Productivity, Building Information Modeling, Computer vision, Modular Construction
License
Resource Type
Date Issued
Degree Level
Degree Name
Degree Field
Degree Grantor
Commencement Year
Advisor
Committee Member
Academic Affiliation
Rights Statement
Related Items
  • Google Scholar
  • Github
  • Researchgate
Funding Statement (additional comments about funding)
  • National Renewable Energy Lab
Publisher
Peer Reviewed
Language

Relationships

Parents:

This work has no parents.

In Collection:

Items

Thumbnail Title Date Uploaded Visibility Actions
file details PanahiRoshan2024.pdf 2024-03-25 Public Download