3D Model Construction Progress Tracking via Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
There is a need to improve the transfer of complex 3D domain, such as a BIM building, into a discrete 2D domain and vice versa, while effectively tracking the construction of objects at a construction site, which are subject to multiple stages and varying lighting conditions, and to provide a cost-effective overview of construction progress.
Innovation Solution
A method utilizing a digital representation of a 3D model with machine learning models to identify objects and their construction stages, involving image capture, orientation, and projection, followed by machine learning-based detection and comparison with a 3D model, allowing for precise progress tracking and visualization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If visual inspection methods are used to monitor construction progress, then cost-effectiveness is improved, but measurement precision and automation extent deteriorate
Solution Approach 1:
The patent replaces manual visual inspection with an automated computer vision system that uses image processing and machine learning algorithms to detect construction objects and track progress. This substitution maintains cost-effectiveness by eliminating manual labor while significantly improving measurement precision through automated object detection and classification.
Solution Approach 2:
The system enables self-service monitoring by automatically capturing images, processing them through machine learning models, and generating progress reports without human intervention. The automated pipeline includes image capture, object detection, stage classification, and progress calculation, allowing the system to monitor itself and provide actionable insights.
2Measurement precision
If multiple images are captured and processed to track construction progress, then measurement precision is improved, but loss of time increases
Solution Approach 1:
The system performs preliminary actions by pre-processing images and pre-training machine learning models with construction stage data before actual progress tracking. The machine learning model is trained in advance to recognize different construction stages, enabling rapid classification during actual monitoring without time-consuming analysis during the tracking process.
Solution Approach 2:
The patent replaces time-consuming manual image analysis with automated machine learning-based object detection and classification. The system processes multiple images simultaneously using parallel computing and optimized algorithms, maintaining high measurement precision while reducing processing time from hours to minutes.
3Productivity
If construction objects are tracked through multiple stages by different craftsmen, then productivity is improved, but device complexity increases
Solution Approach 1:
The patent implements a universal machine learning model that can detect and classify multiple types of construction objects (walls, floors, ceilings, installations) across different construction stages. The single system handles various object types and stages uniformly, improving productivity through standardized tracking while managing complexity through a unified approach rather than separate systems for each object type.
Solution Approach 2:
The system tracks construction progress by monitoring parameter changes in objects, such as completion stage, material installation, and structural configuration. By focusing on key parameters rather than detailed manual inspection of all aspects, the system improves productivity through automated parameter detection while keeping the system relatively simple through targeted monitoring.
4Adaptability or versatility
If images are captured under varying lighting conditions including shadows, then adaptability is improved, but difficulty of detecting and measuring increases
Solution Approach 1:
The patent replaces human visual inspection, which is sensitive to lighting conditions, with machine learning-based image processing that is inherently more robust to varying lighting. The system uses trained models that can recognize construction objects and stages under different lighting conditions, improving adaptability while reducing the difficulty of detection through automated feature extraction and pattern recognition.
Solution Approach 2:
The system performs preliminary training of machine learning models with diverse image data captured under various lighting conditions, including shadows and different times of day. This pre-training enables the model to generalize well and maintain high detection accuracy across varying lighting conditions without requiring real-time adjustments, thereby improving adaptability while keeping the actual measurement process simple.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present invention relates to a method for tracking progress of the construction of objects, in particular walls comprised in a building based on 3D digital representation. Building Information Modeling (BIM) may provide a digital representation of the physical and functional characteristics of a place, such as a building comprising walls and other objects.