Title: Robot Navigation in Buildings: Localizing, Mapping, and Updating the Building Model
Abstract: Simultaneous Localization and Mapping (SLAM) for robot navigation in buildings has a long history and strong track record. While this method commonly relies on LiDaR scans and point clouds, it is typically challenging to obtain the initial map of the building (manual operations). Recent studies are looking at the use of Building Information Models (BIM models). As such models are becoming more mainstream and available in various data formats (e.g. SPF, JSON,RDF), they are an increasingly relevant source to rely on for such robot navigation (localization and mapping). In this presentation, ongoing work at TU Eindhoven on this topic is presented. An indication is given of how such BIM models can be made available as a live data source, e.g. as part of the Building Management System (BMS) or Digital Twin (DT) or a Personalised Building Data Hub (PDH). Relying on this data source, information retrieval from mobile robots is presented, as well as the diverse options to compare with robot telemetric data. As a final part, this presentation shows how this approach allows to also update the live building model, incl. 3D geometric updates, such that a more high-quality and accurate 3D BIM model can be obtained for the same building. This last part is key to ensuring more high-quality lifecycle data management for buildings, which is needed for circular construction, proper long-term maintenance of buildings, and smart building (data) management.