Development of Semantic BIM for Supporting UAV-Based Reality Capture for Facilities Management Applications

对已建设施（如建筑物）的现实捕捉（RC）是开发可用于设施管理的可靠模型的重要一步。传统上，这些模型是基于设施的设计图纸和/或测量数据开发的。但是，由于设施的老化或由于没有记录的维修和翻新活动，现有设施的设计图纸可能会遗漏或不反映设施的实际状况。Pegasus Research & Technologies（PRT）为加拿大航空航天和国防部门提供研究和工程服务，例如使用人工智能进行无人机（UAV）实验的模拟试验台。最近，PRT的服务产品已经发展到包括使用安装在UAV上的激光扫描仪的建筑设施管理应用的RC。然而，将扫描仪收集的数据转换为有用的建筑信息模型（BIM）是一项具有挑战性的任务。本项目假设使用配备激光扫描仪的无人机来收集创建建筑物内部空间3D模型所需的数据。此外，它假设有一个算法forobject识别/分类的基础上收集的传感器数据的无人机。结果是建筑构件（如墙、窗、门等）。已经在语义图中定义了。然而，在这一领域的差距是，目前还没有直接的方式将这种语义地图与BIM联系起来。该项目旨在：（1）开发语义地图和标准BIM类之间的接口;（2）根据收集的数据自动创建BIM模型，以支持后续的建筑检查任务。

Reality capture (RC) of constructed facilities (e.g. buildings) is an important step towards developing reliableas-is models that can be used for facilities management. Traditionally, these models are developed based ondesign drawings and/or surveying data of the facilities. However, design drawings of existing facilities may bemissing or not reflecting the actual conditions of the facilities because of accumulated deterioration or becauseof maintenance and renovation activities that were not documented. On the other hand, conventional surveyingmethods are expensive, time consuming, and may not be safe in some cases (e.g. post-disaster inspection).Pegasus Research & Technologies (PRT) provides research and engineering services to the Canadian aerospaceand defense sector, such as simulation test-bed for Unmanned Aerial Vehicles (UAV) experimentation usingartificial intelligence. Recently, the service offering of PRT has grown to include RC for buildings facilitiesmanagement applications using a laser scanner mounted on the UAV. However, transforming the data collectedby the scanner to a useful Building Information Model (BIM) is a challenging task.This project assumes that a UAV equipped with a laser scanner is used for collecting the data needed forcreating a 3D model of the interior spaces in a building. Furthermore, it assumes that there is an algorithm forobject recognition/classification based on the sensory data collected by the UAV. The results of the recognitionare building components (e.g. wall, window, door, etc.) that are already defined in a semantic map. However, agap in this area is that there is currently no direct way of linking this semantic map with BIM. The project aimsto: (1) develop an interface between the semantic map and standard BIM classes; and (2) automatically create aBIM model based on the collected data in order to support the subsequent building inspection tasks.