Metraipolis › archive › dc_90e04e4e8126

ActSafeGuard: Differentiable and Training-Aligned Constraint Enforcement for Flow-Matching Policies — arXiv:2609.11697

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.11697
License (as archived)
CC0-1.0
Archived
2026-10-16T05:27:13.566Z
SHA-256 of the archived text
a75c6540316174345c8e2bb913d577f8ed047f14b7c29d62834ba7f78da4af91
Stable URL of this page
https://metraipolis.com/documento/dc_90e04e4e8126

How to cite

ActSafeGuard: Differentiable and Training-Aligned Constraint Enforcement for Flow-Matching Policies — arXiv:2609.11697 — arXiv (en), https://arxiv.org/abs/2609.11697 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_90e04e4e8126 (Metraipolis archive dc_90e04e4e8126, sha256 a75c6540316174345c8e2bb913d577f8ed047f14b7c29d62834ba7f78da4af91)
BibTeX
@misc{dc90e04e4e8126,
  title = {ActSafeGuard: Differentiable and Training-Aligned Constraint Enforcement for Flow-Matching Policies — arXiv:2609.11697},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.11697},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_90e04e4e8126; sha256 a75c6540316174345c8e2bb913d577f8ed047f14b7c29d62834ba7f78da4af91},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.11697. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Vision-Language-Action (VLA) and World-Action Models (WAMs) have demonstrated strong capabilities in general-purpose robotic manipulation, yet their generated actions may violate hard physical constraints and therefore be unsafe or infeasible for deployment. Existing safety approaches either optimize statistical safety objectives without deterministic per-step guarantees or correct unsafe actions only during inference, creating a mismatch between policy training and execution. We introduce ActSafeGuard, a differentiable and training-aligned safeguard layer for flow-matching based policies. ActSafeGuard integrates hard action feasibility into policy learning, not merely treating safety as an inference-time external component. Through an analytical ray-scaling operator design, ActSafeGuard enables boundary-aware gradients to guide the model to naturally learn constrained manifolds. Extensive experiments on multiple standard foundation backbones ($\pi_{0.5}$ and Fast-WAM) across various tasks demonstrate that ActSafeGuard consistently achieves a $100\%$ step safety rate while fully preserving or even boosting task success rates, providing a scalable and minimally invasive solution for safe embodied AI deployment.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-11 in cs.AI).
Categorie arXiv: Robotics (cs.RO) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.11697 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-16T05:17:52.187Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-11?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).