PhyScene3D: Physically Consistent 3D Interactive Tabletop Scene Generation

Abstract

Generating physically consistent 3D tabletop scenes is a fundamental yet underexplored problem for interactive and generalist robotic learning. The challenge stems from dense object hierarchies and irregular affordances. Existing methods, ranging from decoupled symbolic solvers to end-to-end regression models, often suffer from error propagation or overfitting to noisy supervision containing widespread physical violations. To address these limitations, we introduce PhyScene3D, a framework that reformulates generation as a Human-Mimetic Constructive Process. The proposed Cognitive Topological Reasoning Chain (CTRC) factorizes scene synthesis into a sequential, anchor-conditioned process under the novel 3D Axis-Aligned Bounding Box (3D AABB)-based placement scheme, thereby imposing a strong structural inductive bias. To address imperfect supervision and physical infeasibility, we introduce Physics-Aware Denoising Alignment (PADA), which integrates a differentiable Signed Distance Field (SDF) with Test-Time Optimization (TTO) to project generated scenes onto a physics-feasible manifold while preserving semantic intent. Experiments demonstrate that PhyScene3D outperforms state-of-the-art approaches in both semantic accuracy and physical validity, achieving a 40\% reduction in collision rate relative to the human-annotated training data.