Orchestrating Spatial Semantics via a Zone-Graph Paradigm for Intricate Indoor Scene Generation

Autonomous 3D indoor scene synthesis breaks down in non-convex rooms with tightly coupled spatial constraints. Data-driven generators lack topological priors for long-horizon planning, while iterative agents fragment semantics and become geometrically brittle. We present \textbf{ZoneMaestro}, a unified framework that shifts the paradigm from object-centric synthesis to Zone-Graph Orchestration. By internalizing a novel zone-based logic, ZoneMaestro translates high-level semantic intent into functional zones and topological constraints, enabling robust adaptation to diverse architectural forms. To support this, we construct \textit{Zone-Scene-10K}, a large-scale dataset enriched with explicit Zone-Graph annotations. We further introduce an \textit{Alternating Alignment Strategy} that cycles between reasoning internalization and Zone-Aware Group Relative Policy Optimization (\textit{Z-GRPO}), effectively reconciling the tension between semantic richness and geometric validity without relying on external physics engines. To rigorously evaluate spatial intelligence beyond convex primitives, we formally define the task of \textbf{Intricate Spatial Orchestration} and release SCALE, a stress-test benchmark for irregular indoor scenarios with complex, dense spatial relations. Extensive experiments demonstrate that ZoneMaestro resolves the density-safety dichotomy, significantly outperforming state-of-the-art baselines in both structural coherence and intent adherence.

Training robots and virtual agents to navigate and interact with indoor environments requires large numbers of realistic 3D room layouts. Today, AI systems can automatically arrange furniture in simple rectangular rooms, but real homes and offices are rarely that simple—they have L-shaped hallways, angled walls, and alcoves that demand careful spatial planning. When current methods encounter these complex floor plans, they produce rooms where furniture pokes through walls, objects pile on top of each other, or large areas are left empty. We address this challenge by teaching an AI model to think about rooms the way an interior designer does: first divide the space into purposeful zones (a dining area, a reading nook, a workspace), then decide how furniture should be arranged within and between those zones. This structured reasoning allows the model to handle irregular room shapes and pack many objects into a scene without physical violations—all in a single generation step rather than repeated trial and error. We also release a large dataset of expert-annotated room layouts and a new benchmark of architecturally complex rooms to support future research. Our system significantly outperforms existing approaches in both automated measurements and human evaluations, bringing AI-generated indoor environments closer to the quality needed for robotics simulation, game design, and computer-aided architecture.