Publication
arXiv
Stage
Preprint
What we read
Summary of the abstract
Authors
Kevin Qu, Tao Sun, Massimiliano Viola, Liyuan Zhu, Zhizhuo Zhou, Sayan Deb Sarkar, Konrad Schindler, Iro Armeni
Universities and research institutions
Not yet supplied in verified metadata; the Brief does not guess.

Institution metadata: OpenAlex record ↗

What the paper reports

The paper introduces FAMOS, a model that aggregates sparse partial point clouds to predict movable parts and joints, using a Multi-state Articulation Transformer, and a data generator to train on diverse synthetic assets.

Why it matters

This abstract-only preprint demonstrates improvements over baselines in controlled settings, but results hinge on synthetic data and abstract objectives.

FAMOS tackles the challenge of modeling articulated objects from sparse monocular observations by predicting both movable-part segmentation and joint parameters from an unordered set of partial point clouds. It uses a Multi-state Articulation Transformer to aggregate cues across observations and handles variable input counts, including a single view.

To train, the authors introduce a procedural data generator that creates self-annotated assets, aiming to broaden scale and diversity beyond existing datasets, and they report improvements over prior feed-forward and optimization-based methods on several benchmarks.

What this does not tell us

This is an abstract-only preprint. The reported results rely on synthetic data and may not fully reflect real-world performance. Further validation on real-world datasets is needed.

Original sources · 1
  1. FAMOS: Feed-Forward 3D Articulation Modeling from Sparse Observations ↗arXiv · 2026-09-17

Check the original paper for its authors, methods, version and access terms.