EAPSI: Construction progress: Monitoring and quality control using smartphones

施工监控为施工从业者--业主、承包商、分包商和贸易商--提供项目控制决策所需的信息。这些决策直接影响到建设项目的整体效率。鉴于全球建筑支出高达数万亿美元，即使是小幅提高能效，也可能为世界带来巨大的成本节约。该项目的目标是通过新的进度监控和质量控制来降低施工成本和交付时间。这些信息将从大量图像和视频流中收集，这些图像和视频流将由智能手机和平板电脑等消费级智能设备提供。这项研究将在韩国首尔国立大学与大规模建筑领域的专家研究员朴元熙博士合作进行。拟议项目的目标是通过计算机视觉和施工管理的进步来自动化大规模建筑施工监控的过程。这项研究将通过提供自动化的3D竣工模型来增强和简化当前建筑工地的进度监控和质量控制任务，这些模型可以与建筑信息模型结合使用，以分析进度和质量。使用廉价的消费级智能设备，施工经理可以轻松收集数据并将输入快速提交到服务器，并通过支持有效项目控制决策的界面可视化结果。这将通过解决计算机视觉和施工管理方面的基本研究挑战来实现，包括通过基于GPU的运动结构架构提高基于图像的三维重建方法的效率和可靠性；从二维图像识别材料属性和几何图形；建立用于外地报告的增强现实可视化语法。该NSF EAPSI奖是与韩国国家研究基金会合作资助的。

Construction monitoring provides construction practitioners -owners, contractors, subcontractors, and tradesmen- with the information need for project control decisions. These decisions directly impact the overall efficiency of a construction project. Given that global construction spending is trillions of dollars, even small gains in efficiency could lead to enormous cost savings to the world. The goal of this project is to reduce both construction cost and delivery time with new progress monitoring and quality control. This information will be gathered from a large set of images and video streams that will be provided by consumer-level smart devices, such as smartphones and tablets. This research will be conducted at Seoul National University in South Korea in collaboration with Dr. Moonseo Park, an expert researcher in large-scale construction. The goal of the proposed project is to automate the process of large-scale building construction monitoring through advances in computer vision and construction management.This research will enhance and simplify current progress monitoring and quality control tasks on construction sites by providing automated 3D as-built models that can be used in conjunction with Building Information Models for analyzing progress and quality. Using inexpensive consumer-level smart devices, construction managers can easily collect data and quickly submit inputs to the server and visualize outcomes with interfaces that enable effective project control decisions. This will be achieved by addressing fundamental underlying research challenges in computer vision and construction management including improving the efficiency and reliability of image-based 3D reconstruction method through GPU-based architecture for structure-from-motion; recognizing material properties as well as geometry from 2D images; and establishing a grammar in augmented reality visualization for field reporting. This NSF EAPSI award is funded in collaboration with National Research Foundation of Korea.