PD$^{2}$GS: Part-Level Decoupling and Continuous Deformation of Articulated Objects via Gaussian Splatting

TL;DR

PD$^{2}$GS introduces a novel framework for modeling articulated objects by learning a shared Gaussian field and representing interaction states as continuous deformations, enabling accurate part-level decoupling and smooth control without manual supervision.

Contribution

It proposes a unified approach that encodes geometry and kinematics jointly, refines part boundaries with vision priors, and supports continuous control and accurate modeling of articulated objects.

Findings

Outperforms prior methods in geometric accuracy

Achieves superior kinematic modeling and control consistency

Demonstrates effectiveness on both synthetic and real datasets

Abstract

Articulated objects are ubiquitous and important in robotics, AR/VR, and digital twins. Most self-supervised methods for articulated object modeling reconstruct discrete interaction states and relate them via cross-state geometric consistency, yielding representational fragmentation and drift that hinder smooth control of articulated configurations. We introduce PD$^{2}$GS, a novel framework that learns a shared canonical Gaussian field and models the arbitrary interaction state as its continuous deformation, jointly encoding geometry and kinematics. By associating each interaction state with a latent code and refining part boundaries using generic vision priors, PD$^{2}$GS enables accurate and reliable part-level decoupling while enforcing mutual exclusivity between parts and preserving scene-level coherence. This unified formulation supports part-aware reconstruction, fine-grained continuous control, and accurate kinematic modeling, all without manual supervision. To assess realism and generalization, we release RS-Art, a real-to-sim RGB-D dataset aligned with reverse-engineered 3D models, supporting real-world evaluation. Extensive experiments demonstrate that PD$^{2}$GS surpasses prior methods in geometric and kinematic accuracy, and in consistency under continuous control, both on synthetic and real data.