Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces a novel rest-state formulation for articulated object reconstruction that enables the recovery of 3D geometry and kinematic structures from a single closed configuration, addressing the limitations of existing methods that rely on multiple observable motion states. By leveraging an explicit mesh for cross-model verification and fusing outputs from vision-language and segmentation models, the framework compensates for the lack of motion cues through geometry, semantics, and motion priors. The proposed approach not only achieves accurate part decomposition but also demonstrates physically plausible articulation, outperforming traditional motion-observing and generation-based baselines.
Recovering articulated object structures from a single closed configuration could revolutionize how we build interactive digital twins.
Building interactive digital twins requires recovering both 3D geometry and the kinematic structures that govern how objects articulate. Yet existing methods for articulated object reconstruction require explicitly observable motion from multiple articulation states. We introduce a rest-state formulation that reconstructs articulated objects from a single closed configuration, an inherently ill-posed setting where geometry, semantics, and motion priors compensate for the absence of motion cues. Our framework adopts an explicit mesh as an intermediate representation for cross-model verification and fusion, reconciling noisy outputs from vision-language and segmentation models into spatially consistent part structures. To estimate joint parameters without observed motion, we use a video diffusion model to synthesize articulation hypotheses and validate them through geometric consistency. Our approach achieves accurate part decomposition and physically plausible articulation, performing competitively with motion-observing reconstruction-based, generation-based, and modular pretrained-model baselines.