Automated Vision-Based Sensing for Site Operations Analysis

本研究旨在证明，为了再现与建筑工地相关的日常工作流程活动，可以可靠、自动地跟踪工作进度和使用图像（视频和/或延时）的多种资源。测量涉及工人、大型机器和材料的施工现场活动的进度的任务通常是一个主观的、密集的人工过程，容易出错，而且在实际操作中经常过时。演示一个主动视觉系统可以有效地分析和评估工作现场的进度，将通过减少花费在监控和解释项目状态和性能上的时间来帮助项目经理，从而增加对成本和进度控制的关注。通过使项目管理人员和工作人员更加了解他们的项目和工作环境的性能状态，可以预见到行业的潜在节省。轨道数据将被解释并用于提供对工作地点时空演变的理解，以自动生成有关工作地点操作的知识。在基于信息的框架中，需要花费大量精力获取和解释信息。在以知识为基础的框架中，工作分配到根据解释的信息做出决策。如果成功，本研究将使施工作业的审查和管理从信息化转变为知识化，从而节省人力资源，提高决策效率。这项研究在建筑之外具有更广泛的吸引力。研究领域结合或需要基于视觉的传感，不同的资源（人，小型到重型机械，货物等），以及为操作和活动的意识处理视觉数据是额外的研究领域。例子包括机场地面作业和采矿作业。在机器学习和计算机视觉领域也有望有所贡献。拟议的研究将通过实现对站点资源的自动监测和跟踪，影响对站点操作的研究。基于视频的监控和处理算法提供了一种非侵入式、简单和快速的机制，可以生成大量的操作信息和知识，这些信息和知识一旦提供，就可以对目前无法实现的施工操作进行查询。从长远来看，这项研究将通过实现更严格的控制和更高的效率，为项目管理提供有价值的帮助。通过使项目管理人员和工作人员更加了解他们的项目和工作环境的性能状态，可以预见到建筑和其他行业的潜在节省。这项研究还将积极包括并推动下一代工程师（土木，电气和计算工程）和建筑劳动力的教育。这项研究有一个专门的外展计划，让对先进的硬件和软件技术感兴趣的高中生和行业专业人士广泛参与这项研究

This research seeks to prove that it is possible to reliably and automatically track work progress and multiple resources with images (video and/or time-lapse) in order to reproduce the daily workflow activities associated to a construction worksite. The task of measuring the progress of construction site activities that involve workers, large machines, and materials, has often been a subjective and intensive manual process that is prone to error and, in real operations, frequently out-of-date. Demonstrating that an active vision system can effectively analyze and assess work-site progress will assist project managers by reducing the time spent monitoring and interpreting project status and performance, thus enabling increased attention to control of cost and schedule. By making project management and workforce more aware of the performance status of their project and their work environment, potential savings to the industry are envisioned. The track data will be interpreted and used to provide understanding of the spatio-temporal evolution of a worksite for automatically generating knowledge about worksite operations. In an information-based framework, much effort is spent acquiring and interpreting information. In a knowledge-based framework, efforts are allocated to making decisions based on the interpreted information.If successful, this research will transform the review and management of construction operations from being information-based to knowledge-based, thus saving human resources and improving decision effectiveness. This research has broader appeal beyond construction. Research domains incorporating or requiring vision-based sensing, diverse resources (people, small to heavy machinery, goods, etc.), and processing of the visual data for awareness of operations and activities are additional investigation domains. Examples include airport ground operations and mining operations. Contributions are also expected in the fields of machine learning and computer vision. The proposed research will impact research into site operations by enabling the automated monitoring and tracking of site resources. Video-based monitoring and processing algorithms provide a non-intrusive, easy, and, rapid mechanism for generating a body of operational information and knowledge which, when made available, will enable inquiry into construction operations that is currently not possible. Longer term, this research will serve as a valuable aid to project management by enabling tighter control and greater efficiency. By making project management and workforce more aware of the performance status of their project and their work environment, potential savings to the construction and other industries are envisioned. This research will also actively include and drive the education of the next generation of engineers (civil, electrical, and computational engineering) and construction labor pool. The research has a dedicated outreach plan to involve in this research a broad spectrum of students from high schools and industry professionals who are interested in advanced hard- and software technology