NSF Convergence Accelerator Track D: Rapid Development of Intelligent, Built Environment Geo-Databases Using AI and Data-Driven Models

NSF融合加速器支持以使用为灵感，以团队为基础，多学科的努力，解决国家重要性的挑战，并将在不久的将来产生对社会有价值的可交付成果。“融合加速器”第一阶段项目的更广泛影响和潜在社会效益是实现建筑环境研究界和工业界所认为的圣杯。基础设施设计师和规划者越来越需要建筑环境的详细三维（3D）地理空间数据库，以便考虑人类尺度感知，这在今天经常被忽视，但对于改善城市以更好地满足其公民和其他新兴技术（如自动驾驶汽车）的需求至关重要。通过将原始数据转化为可操作的信息，应用人工智能（AI）快速创建安全智能的建筑环境3D模型，包括水平（如高速公路、桥梁）和垂直（如建筑物、工厂）设施，对社会的影响不容小觑。然而，迄今为止，即使对最先进的城市来说，组装和维护这些数据库也过于昂贵，这在很大程度上是由于人工特征提取的成本。准确可靠的建筑环境数据对于社区的福祉至关重要，因为此类数据用于各种目的，包括应急准备、资产运营、维护、公共安全等。我们将与业界合作伙伴组成一个融合的创新团队，以确保通过本研究开发的知识有效地转化为实践的许多方面。从一开始，通过项目的两个阶段，团队通过团队建设和与各种利益相关者的有意接触采取步骤，以扩大拟议创新的潜在影响范围。具体而言，研究和实施活动的设计将始终包括对话，并邀请广泛的利益相关者提供意见，有意寻求在技术实施努力中传统上被遗漏或忽视的公众部分的参与。这项研究还实施了计算机视觉和地理信息学方面的劳动力发展活动，目前在就业需求和适当技术人员的劳动力之间存在很大差距，特别是来自代表性不足的背景。目前的工作流程和程序，以开发建筑环境的3D模型需要大量的手工工作。这些自动化的过程仅限于小型数据集，这些数据集不能代表当前的点云以及建筑信息建模（BIM）所获取或需要的其他数据。它们在可以建模的对象类型方面也受到限制。为此，跨学科研究团队将与利益相关者合作，开发更全面的扫描到bim流程。该项目的第一阶段有两个主要目标：(1)通过编译大量带有注释的点云扫描和相应的BIM模型的基准数据集，提供扫描到BIM的验证工具，创建一个原型验证服务器，其中包含与利益相关者感兴趣的参数相关的指标（例如，评估建模门宽度的准确性，这对ADA合规性评估很重要，或评估城市更新、再开发项目的这些模型）；(2)开发一个原型工具来实现一个整体的scan -to-BIM框架，从扫描数据中快速可靠地生成BIM模型，该模型不仅可以用来促进基准数据集的开发，而且可以被利益相关者使用。基于这些挑战的研究结果，项目团队计划为Scan-to-BIM构建一个全面的基于云的服务，该服务将被部署到建筑/工程/建筑（AEC）社区，并最终服务于公众，因为这些模型可以降低纳税人资助的公共基础设施的建设或翻新项目成本。用户将能够根据特定应用程序和用例的性能，为Scan-to-BIM框架的关键阶段选择所需的算法。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The NSF Convergence Accelerator supports use-inspired, team-based, multidisciplinary efforts that address challenges of national importance and will produce deliverables of value to society in the near future. The broader impact and potential societal benefit of this Convergence Accelerator Phase I project is to achieve what is seen as the Holy Grail by the Built Environment research community and industry. Infrastructure designers and planners increasingly require detailed three dimensional (3D) geospatial databases of the Built Environment, in order to consider human scale perceptions, which today are often overlooked but critical to improving cities in order to better meet the needs of its citizens and other emerging technologies, such as autonomous vehicles. The impact on society of applying Artificial Intelligence (AI) to rapidly create secure and intelligent 3D models of the Built Environment, including horizontal (e.g., highway, bridges) and vertical (e.g., buildings, plants) facilities, by turning raw data into actionable information cannot be overstated. However, to-date, assembling and maintaining those databases has been too expensive for even the most progressive cities due in large part to the cost of manual feature extraction. Built Environment data that is accurate and reliable is critical for the well-being of communities as such data is used for a variety of purposes including emergency preparedness, asset operations, maintenance, public safety, and more. A convergent, innovative team will be formed with industry partners to ensure that the knowledge developed through this research effectively transitions into many aspects of practice. From the outset, and through both phases of the project, the team is taking steps through team building and intentional engagement with a variety of stakeholders to broaden the scope of potential impacts of the proposed innovation. Specifically, the research and implementation activities will be designed to consistently include dialogue and invite input from a broad range of interests, intentionally seeking involvement with segments of the public that are traditionally left out or neglected in technology implementation endeavors. This research also implements activities for workforce development in computer vision and geomatics, which currently has a large gap between employment needs and a workforce of appropriately skilled personnel, particularly from underrepresented backgrounds. Current workflows and procedures to develop 3D models of the Built Environment require substantial manual effort. Those processes that are automated are limited to small datasets that are not representative of the current point clouds and other data being acquired or needed for Building Information Modeling (BIM). They are also limited in the types of objects that can be modeled. To this end, the interdisciplinary research team will work with stakeholders to develop a more holistic Scan-to-BIM process. Phase I of this project has two primary goals: (1) provide a scan-to-BIM validation tool by compiling a sizable collection of benchmark datasets with annotated point cloud scans and corresponding BIM models, creating a prototype validation server with metrics related to parameters of interest to stakeholders (e.g., evaluate the accuracy of modeled door widths, which are important to ADA compliance assessment, or evaluate these models for urban renewal, redevelopment projects), and create and host a Scan-to-BIM challenge for researchers all over the world to participate, and (2) develop a prototype tool to implement a holistic Scan-to-BIM framework to rapidly and reliably generate BIM models from scan data that can be used not only to facilitate the development of the benchmark datasets but also used by stakeholders. Based on the research resulting from these challenges, the project team plans to build a comprehensive cloud-based service for Scan-to-BIM, which will be deployed to serve the Architectural/Engineering/Construction (AEC) community, and ultimately the public in general as these models can decrease construction or renovation project costs of public infrastructure funded with taxpayer’s money. Users would be able to select desired algorithms for key stages of the Scan-to-BIM framework based on their performance for specific applications and use cases.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.