Space Syntax-guided Post-training for Residential Floor Plan Generation

2026-02-25

AI Summary

This work addresses the common oversight in existing residential floor plan generation models, which often neglect key architectural priors such as the configurational dominance and connectivity of public spaces (e.g., living rooms, foyers). To remedy this, we propose SSPT, a post-training paradigm that integrates space syntax theory into data-driven generation in a scalable manner. Our approach constructs a rectangle-to-space graph via maximal rectangle decomposition, computes integration values using door-mediated adjacency, and employs a non-differentiable space syntax Oracle to guide generation. We design an efficient Proximal Policy Optimization (PPO)-based reinforcement learning strategy for post-training refinement and introduce SSPT-Bench, a dedicated evaluation benchmark. Experiments demonstrate that our method significantly enhances the dominance of public spaces and the clarity of functional hierarchies in generated layouts, with the PPO strategy outperforming baselines in both efficiency and stability.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Search and Optimization: Sampling/Simulation-based Search
• Natural Language Processing: Generation

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

Pre-trained generative models for residential floor plans are typically optimized to fit large-scale data distributions, which can under-emphasize critical architectural priors such as the configurational dominance and connectivity of domestic public spaces (e.g., living rooms and foyers). This paper proposes Space Syntax-guided Post-training (SSPT), a post-training paradigm that explicitly injects space syntax knowledge into floor plan generation via a non-differentiable oracle. The oracle converts RPLAN-style layouts into rectangle-space graphs through greedy maximal-rectangle decomposition and door-mediated adjacency construction, and then computes integration-based measurements to quantify public space dominance and functional hierarchy. To enable consistent evaluation and diagnosis, we further introduce SSPT-Bench (Eval-8), an out-of-distribution benchmark that post-trains models using conditions capped at $\leq 7$ rooms while evaluating on 8-room programs, together with a unified metric suite for dominance, stability, and profile alignment. SSPT is instantiated with two strategies: (i) iterative retraining via space-syntax filtering and diffusion fine-tuning, and (ii) reinforcement learning via PPO with space-syntax rewards. Experiments show that both strategies improve public-space dominance and restore clearer functional hierarchy compared to distribution-fitted baselines, while PPO achieves stronger gains with substantially higher compute efficiency and reduced variance. SSPT provides a scalable pathway for integrating architectural theory into data-driven plan generation and is compatible with other generative backbones given a post-hoc evaluation oracle.

Problem

• residential floor plan generation
• space syntax
• public space dominance
• functional hierarchy
• architectural priors

Innovation

• Space Syntax
• Post-training
• Floor Plan Generation
• Reinforcement Learning
• Architectural Prior