Reasoning Spatial Relationships in Building Information Models using Voxels

This paper investigates a voxel-based approach to automate the space construction of an ill-defined building information model, namely, a building model without specific spatial definitions. The objective is to provide a simplified representation through clustering voxels to reconstruct spaces, with which a spatial topological algorithm is designed to infer the implicit connectivity. This approach is treated as the first step to automate building information exchange for building performance simulation and knowledge-intensive reasoning.