SIMART: Decomposing Monolithic Meshes into Sim-ready Articulated Assets via MLLM

TL;DR

SIMART is a novel framework that decomposes monolithic 3D meshes into articulated assets suitable for simulation, using a sparse tokenization approach with MLLMs to improve scalability and accuracy.

Contribution

It introduces a Sparse 3D VQ-VAE to reduce token counts and jointly performs part decomposition and kinematic prediction in a unified MLLM framework.

Findings

Achieves 70% reduction in token count compared to dense voxel methods.

Sets new state-of-the-art on PartNet-Mobility and AIGC datasets.

Enables realistic physics-based robotic simulation.

Abstract

High-quality articulated 3D assets are indispensable for embodied AI and physical simulation, yet 3D generation still focuses on static meshes, leaving a gap in "sim-ready" interactive objects. Most recent articulated object creation methods rely on multi-stage pipelines that accumulate errors across decoupled modules. Alternatively, unified MLLMs offer a single-stage path to joint static asset understanding and sim-ready asset generation. However dense voxel-based 3D tokenization yields long 3D token sequences and high memory overhead, limiting scalability to complex articulated objects. To address this, we propose SIMART, a unified MLLM framework that jointly performs part-level decomposition and kinematic prediction. By introducing a Sparse 3D VQ-VAE, SIMART reduces token counts by 70% vs. dense voxel tokens, enabling high-fidelity multi-part assemblies. SIMART achieves state-of-the-art performance on PartNet-Mobility and in-the-wild AIGC datasets, and enables physics-based robotic simulation.