SBIR Phase I: Analysis of Progress Photos for Indoor Construction Progress Monitoring

SBIR 第一阶段：室内施工进度监控的进度照片分析

• 批准号: 1819248
• 负责人: Derek Hoiem
• 金额: $ 22.5万
• 依托单位: RECONSTRUCT INC
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2018
• 资助国家: 美国
• 起止时间: 2018-06-15 至 2018-11-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1819248&HistoricalAwards=false
• 关键词: SBIR; Phase; Analysis; Progress; Photos

The broader impact/commercial potential of this Small Business Innovation Research (SBIR) Phase I project is that it will save billions of dollars by anticipating construction delays and streamlining coordination to prevent them. Lower costs will help upgrade our nation's infrastructure, which is a critical national priority. Many construction companies strongly want a solution to indoor progress monitoring. Laser scanning is too expensive and slow, and photography services can cost hundreds of thousands of dollars for large projects and introduce logistical challenges. The proposed research would simplify the workflow for progress monitoring of interiors by automatically registering 360 degree photos and video and aligning them to building information models (BIM), providing a cheap, quick, and effective solution for daily progress monitoring.This Small Business Innovation Research (SBIR) Phase I project aims to localize images, reconstruct 3D models, align them to floorplans or 3D BIM, and use the aligned models to provide actionable data to construction managers. A main challenge is to robustly solve for camera pose using structure-from-motion in indoor scenes that are unfinished and contain textureless and reflective surfaces. A second challenge is to automatically or semi-automatically register the models to 2D or 3D plans, made more difficult by the fact that the site is incomplete and constantly changing. A third challenge is to create interfaces that project personnel can use to perform visual inspection, progress monitoring, requests for information, and other supervision and coordination tasks. The project will investigate the use of tags (i.e., markers) to improve robustness of 3D reconstruction and to perform registration without requiring the 3D positions of tags to be known in advance. Thus, the project addresses major unsolved problems in computer vision, robotics, and their application to construction management.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

该小企业创新研究 (SBIR) 第一阶段项目的更广泛影响/商业潜力在于，它将通过预测施工延误并简化协调来防止延误，从而节省数十亿美元。降低成本将有助于升级我们国家的基础设施，这是国家的一项重要优先事项。许多建筑公司强烈希望有一个室内进度监控的解决方案。激光扫描成本太高且速度太慢，大型项目的摄影服务可能花费数十万美元，并带来后勤挑战。拟议的研究将通过自动注册 360 度照片和视频并将其与建筑信息模型 (BIM) 对齐来简化室内进度监控的工作流程，为日常进度监控提供廉价、快速且有效的解决方案。这项小型企业创新研究(SBIR) 第一阶段项目旨在本地化图像、重建 3D 模型、将其与平面图或 3D BIM 对齐，并使用对齐的模型为施工经理提供可操作的数据。主要挑战是在未完成且包含无纹理和反射表面的室内场景中使用运动结构稳健地解决相机姿势问题。第二个挑战是自动或半自动将模型注册到 2D 或 3D 平面图，由于站点不完整且不断变化，这一挑战变得更加困难。第三个挑战是创建项目人员可以用来执行目视检查、进度监控、信息请求以及其他监督和协调任务的界面。该项目将研究标签（即标记）的使用，以提高 3D 重建的鲁棒性，并在不需要提前知道标签 3D 位置的情况下执行配准。因此，该项目解决了计算机视觉、机器人技术及其在施工管理中的应用方面尚未解决的主要问题。该奖项反映了 NSF 的法定使命，并通过使用基金会的智力价值和更广泛的影响审查标准进行评估，被认为值得支持。