From Sketching to BIM – Workflow for the generation of IFC-based BIM models from 4D semantic sketches

Abstract

The preliminary development of highly detailed BIM models lacks eligible design methods in the conceptual design stage. Initial design solutions cannot be shared among collaborative and consecutive design parties without relying on error-prone briefings and data exchanges. As advances in Artificial Intelligence are increasingly being used for BIM's big data challenges, this paper presents a workflow based on a small scope of IFC definitions for the generation of conceptual BIM models from 4D semantic sketches, using a machine learning based pipeline. The workflow offers a user-friendly architectural design perspective, while preserving information more reliably throughout the entire life cycle of a BIM model.