DT-PoseFormer: A Digital Twin-enabled Transformer Network for Precise Pose Estimation and Trajectory Prediction of MiC Modules
Author(s)
Han, Yujie
Zhao, Zhiheng
Advisor(s)
Editor(s)
Collections
Supplementary to:
Abstract
Modular integrated construction (MiC) is a game-changing disruptively-innovative approach, which improves quality, simplifies management and reduces construction time due to its off-site production and on-site installation characteristics. However, as one of the most sophisticated processes of MiC, on-site assembly still faces: 1).insufficient interoperability; 2). low-efficient module alignment; 3). highly susceptible safety incidents. To address these challenges, our contributions are: A digital twin (DT)-based real-time module tracking framework for real-time monitoring and control of MiC assembly to fulfill sufficient interoperability. A transformer-enabled (DT-PosFormer) network for real-time spatial-temporal module data analysis to automatic pose estimation. A trajectory prediction of MiC assembly alignment to prevent potential structural collision and misaligned stacking.
Sponsor
Date
2024-05
Extent
Resource Type
Text
Resource Subtype
Poster