Moving from targeted acquisition to urban area modelling — increasing the scale of point cloud processing

Abstract

Handling real world, acquired point cloud data within a sector such as architecture, engineering, and construction (AEC) is currently a difficult task. A highly desirable, future goal is to fully automate the scan-to-BIM process which at this time has a high dependency on manual work effort. Improvements within this workflow will speed up the production of detailed 3d building models and reduce associated costs. By increasing the level of automation in the scan-to-BIM process it becomes possible to speculate the expansion of the typical use case from a single structure, targeted acquisition towards urban area data collection and modelling. The scale and characteristic differences of an urban area point cloud dataset and that of a single structure create opportunities to validate the applicability of novel analytical approaches to process automation. A decrease in process complexity could be achieved by reducing both the depth of prerequisite knowledge and the level of intervention expected from an operator by a modelling platform. This would also provide an alternative perspective and an opportunity to model operator tasks at a higher, more abstract level. There lacks a completeness of modern documentation within preexisting civil structures. Building information modelling of the as-built condition can reduce overheads associated with key areas such as collaboration, maintenance, and future modifications. However, it is often made more difficult by a lack of accurate documentation due to the age of the building. A common trend that can be observed in countries such as the United Kingdom is that much of future building stock is already standing. Therefore, it is apparent the need to record the detailed, as-built condition of structures essentially from scratch will not resolve itself in the near future. This paper will overview a case study of an urban area modelling conducted in Ebbw Vale, Wales and introduces an abstract scan-to-BIM process automation methodology. This will be supported by a review of a selection of applied research literature. This paper is part of the early development stages of a point cloud processing platform, pcl_toolkit. The command line software aims to simplify approaches commonly associated with various point cloud processing tasks and provide a foundation for rapid development in the near future.