About this paper
SALA aligns latent actions across human and robot videos by using language-based semantic similarities to form a shared latent action space that supervises embodied foundation model training and improves robot task performance.
Embodied foundation models hinge on diverse robot interaction data, which are costly to collect. Egocentric human videos offer a complementary source of experience, and latent action models (LAMs) provide a popular means of leveraging these videos for policy supervision. Compared with world-action models (WAMs), LAMs can offer greater training efficiency and potentially lower inference costs by avoiding explicit pixel prediction during policy learning and execution. However, most existing LAMs learn latent actions solely through video reconstruction, which can cause their representations to cluster by visual appearance rather than by the underlying actions. As a result, they struggle to…
embodied foundation modelslatent action models (LAMs)world-action models (WAMs)policy supervisionvideo reconstructionlatent actionssemantic correspondencecross-embodiment correspondencesoft neighborhood structuresshared latent action space
- Upvotes
- 8
- Authors
- 4
- Published
- Nov 15, 2026
Summary and abstract excerpt via Hugging Face Papers ↗ · arXiv.
Trending history
SALA: Semantically Aligned Latent Actions Enable Dual-Axis Scaling of Embodied Foundation Models was first seen on Hugging Face · Daily papers on Nov 17, 2026. Since then it has appeared in 15 checks on 2 different days. Its best position was #11 of 30, reached on Nov 17, 21:28 UTC. It dropped off the list after Nov 18, 13:28 UTC.