ArtGS: Building Interactable Replicas of Complex Articulated Objects via Gaussian Splatting

Building interactable replicas of articulated objects is a key challenge in computer vision. Existing methods often fail to effectively integrate information across different object states, limiting the accuracy of part-mesh reconstruction and part dynamics modeling, particularly for complex multi-part articulated objects. We introduce ArtGS, a novel approach that leverages 3D Gaussians as a flexible and efficient representation to address these issues. Our method incorporates canonical Gaussians with coarse-to-fine initialization and updates for aligning articulated part information across different object states, and employs a skinning-inspired part dynamics modeling module to improve both part-mesh reconstruction and articulation learning. Extensive experiments on both synthetic and real-world datasets, including a new benchmark for complex multi-part objects, demonstrate that ArtGS achieves state-of-the-art performance in joint parameter estimation and part mesh reconstruction. Our approach significantly improves reconstruction quality and efficiency, especially for multi-part articulated objects. Additionally, we provide comprehensive analyses of our design choices, validating the effectiveness of each component to highlight potential areas for future improvement. Our work is made publicly available at: https://articulate-gs.github.io. Articulated objects, central to everyday human-environment interactions, have become a key focus in computer vision research (Yang et al., 2023a; Weng et al., 2024; Luo et al., 2025; Liu et al., 2024; Deng et al., 2024). As we advance towards more sophisticated robotic systems and immersive virtual environments, there is a growing need for improved and efficient modeling techniques for the reconstruction of articulated objects. The problem of reconstructing articulated objects has been extensively studied (Liu et al., 2023a;b; Weng et al., 2024; Deng et al., 2024; Yang et al., 2023a), with a key challenge being the learning of object geometry when only partial views of the object are available at any given state.