Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

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

Zhuoyang Jiang, Dongqing Zhang

TL;DR

Experiments show that 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 and reinforcement learning via PPO with space-syntax rewards.

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.

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

TL;DR

Experiments show that 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 and reinforcement learning via PPO with space-syntax rewards.

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 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.
Paper Structure (67 sections, 25 equations, 7 figures, 6 tables)

This paper contains 67 sections, 25 equations, 7 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Literature Review
  3. Interdisciplinary Task Formulations for Generative Residential Layouts
  4. Formal Task Definition and Complexity
  5. Methodological Evolution of AI-based Floorplan Generation
  6. Data Foundations and Benchmarking
  7. Regional and Scale Taxonomy
  8. Modality and Evaluation Metrics
  9. Architectural Criteria and Spatial Logic for Plan Assessment
  10. Evaluation and Quality Assessment of Floor Plans
  11. Space Syntax as a Computational Prior for Functional Alignment
  12. Advanced AI Paradigms for Knowledge-Guided Generative Design
  13. Knowledge-guided Generation Systems
  14. (1) Conditional generation.
  15. (2) Evolutionary algorithms and black-box optimization.
  16. ...and 52 more sections

Figures (7)

  • Figure 1: Overview Framework
  • Figure 2: Overall relative integration pattern across room types
  • Figure 3: Dataset-specific room-type coverage and labeling differences
  • Figure 4: Comparison of Relative Integration Profiles.
  • Figure 5: SSPT-Bench (Eval-8) median trends over post-training iterations. SSPT-PPO consistently improves public-space dominance and living-room advantage, while producing a notably narrower (more stable) distribution for living-room relative integration.
  • ...and 2 more figures