Higharc wins EUVIP industrial paper award for BIM-native layout research

Higharc received the Best Industrial Paper Award at EUVIP 2026 in Luxembourg for research on generating room layouts inside BIM constraints. The work points to a faster path for AI-assisted architectural design by preserving building geometry in the model’s input data.

Why it matters: - Higharc’s research targets a core bottleneck in AI-assisted architecture: generating room layouts that are actually usable inside building information modeling, or BIM, workflows. - The work could improve how software places furniture, fixtures and casework within existing room geometry while respecting walls, doors, windows and other constraints. - The research also fed into Higharc’s Generative Building Model, which powers generative design in the company’s platform.

What happened: - Higharc said its paper, "BIM-Native Tokenization for Constraint-Aware Room Layout Synthesis," won the Best Industrial Paper Award at the 14th European Conference on Visual Information Processing, or EUVIP 2026, in Luxembourg. - Manuel Rodríguez Ladrón de Guevara, a senior machine learning engineer at Higharc, accepted the award on Oct. 1, 2026, on behalf of the research team. - The paper was authored by Rodríguez Ladrón de Guevara, Jinmo Rhee, Ardavan Bidgoli, Vaidas Razgaitis and Michael Bergin.

The details: - The research focuses on how to represent building geometry and constraints so a machine learning model can generate designs that fit within BIM environments. - The objective was to automatically arrange furniture, fixtures and casework within existing BIM room models. - The team needed the model to understand room geometry, wall relationships, openings and placement rules that shape how spaces function. - Rodríguez Ladrón de Guevara said the main challenge was representation, not model design, because the team had to encode rooms in a way a designer could actually use. - Higharc said the team spent about a year developing the research results. - The paper introduces the Small Building Model, or SBM, a domain-specific Transformer model built around a BIM-native tokenization method. - The researchers encoded rooms as structured sequences of building elements. - The representation preserves categorical and continuous attributes, including wall types, dimensions, positions and spatial relationships. - The model can analyze and retrieve similar room configurations. - The model can also generate layouts by predicting the placement of room contents within an existing architectural envelope. - In controlled evaluations, the model showed stronger inventory coverage and navigability than two specialized layout-generation baselines trained on the same data. - The results support the researchers’ conclusion that preserving architectural information in the input representation is important for generating usable layouts. - Higharc said the work helped inform the research behind its Generative Building Model, which uses structured building data to generate interior layouts within existing BIM room geometry. - The Generative Building Model accounts for measured walls, doors, windows and relationships between building elements.

Between the lines: - The award gives outside validation to a technical approach that treats building structure as a first-class input for AI, rather than as background context. - The research suggests that better data representation may matter as much as, or more than, raw model complexity in architecture-focused machine learning. - Rodríguez Ladrón de Guevara said the recognition highlights housing as a major but underexplored area for modern AI and could encourage more researchers to work on the field’s technical challenges.

What's next: - Higharc has made the research available at the project site and the paper on arXiv. - The company is likely to continue building on the same underlying research for generative design inside its platform. - Higharc will also keep positioning its platform around design-through-construction workflows for homebuilders and supply chain partners.

The bottom line: - Higharc’s award-winning paper shows how BIM-native data representation could make AI-generated room layouts more practical for real-world housing design.