Unified Vector Floorplan Generation via Markup Representation

2026-04-06

AI Summary

Existing methods for residential floorplan generation struggle to simultaneously ensure functionality and feasibility while flexibly accommodating heterogeneous constraints—such as lot boundaries and room adjacency graphs—and preserving design diversity. To address this, this work proposes the Floorplan Markup Language (FML), a unified structured representation that encodes floorplans as sequences. Building upon FML, we introduce FMLM, a Transformer-based model that reframes multi-condition floorplan generation as a sequential next-token prediction task. This approach achieves the first end-to-end synthesis of vectorized floorplans, and with a single model trained on the RPLAN dataset, it surpasses multiple specialized state-of-the-art methods, significantly improving fidelity, functional correctness, and generalization capability of the generated layouts.

Abstract

Automatic residential floorplan generation has long been a central challenge bridging architecture and computer graphics, aiming to make spatial design more efficient and accessible. While early methods based on constraint satisfaction or combinatorial optimization ensure feasibility, they lack diversity and flexibility. Recent generative models achieve promising results but struggle to generalize across heterogeneous conditional tasks, such as generation from site boundaries, room adjacency graphs, or partial layouts, due to their suboptimal representations. To address this gap, we introduce Floorplan Markup Language (FML), a general representation that encodes floorplan information within a single structured grammar, which casts the entire floorplan generation problem into a next token prediction task. Leveraging FML, we develop a transformer-based generative model, FMLM, capable of producing high-fidelity and functional floorplans under diverse conditions. Comprehensive experiments on the RPLAN dataset demonstrate that FMLM, despite being a single model, surpasses the previous task-specific state-of-the-art methods.

Problem

• floorplan generation
• conditional generation
• representation learning
• generalization
• residential design

Innovation

• Floorplan Markup Language
• Unified Representation
• Transformer-based Generation
• Conditional Floorplan Synthesis
• Next-token Prediction