Metraipolis › archive › dc_d7de09aeb4e6

Turning Safety into Competence: Minimally Exploitable Robot Policies via Safety-Filtered Reinforcement Learning — arXiv:2609.27312

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.27312
License (as archived): CC0-1.0
Archived: 2026-10-08T09:12:26.494Z
SHA-256 of the archived text: e7d23a159765443788904d4840d36d949e95f2304cb84a07f575ed900b99a8e5
Stable URL of this page: https://metraipolis.com/documento/dc_d7de09aeb4e6

How to cite

Turning Safety into Competence: Minimally Exploitable Robot Policies via Safety-Filtered Reinforcement Learning — arXiv:2609.27312 — arXiv (en), https://arxiv.org/abs/2609.27312 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d7de09aeb4e6 (Metraipolis archive dc_d7de09aeb4e6, sha256 e7d23a159765443788904d4840d36d949e95f2304cb84a07f575ed900b99a8e5)

BibTeX

@misc{dcd7de09aeb4e6,
  title = {Turning Safety into Competence: Minimally Exploitable Robot Policies via Safety-Filtered Reinforcement Learning — arXiv:2609.27312},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.27312},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d7de09aeb4e6; sha256 e7d23a159765443788904d4840d36d949e95f2304cb84a07f575ed900b99a8e5},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.27312. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Robots deployed for competitive tasks must outmaneuver their opponents without sacrificing safety. Existing approaches, including safe reinforcement learning (RL), train a single policy to achieve task success and avoid failures simultaneously. This coupling can complicate training and leave the learned policy exploitable by deliberate attacks. We propose Safety to Competence (S2C), a two-stage RL framework that separates safety synthesis from competitive task learning. We formulate competitive interactions as safety-critical Markov games and prove that perfect filtering preserves policy non-exploitability when all players commit to safe maneuvers. S2C learns a robust safety filter via adversarial RL, embeds it in the environment during task policy training, and retains the same filter at deployment. In simulated touchdown games, S2C outperforms eight safe RL baselines, achieving the highest win rate and Elo rating, and the lowest exploitability. Hardware stress tests against a human opponent confirm S2C's competence.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-24 in cs.AI).
Categorie arXiv: Robotics (cs.RO) ; Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Systems and Control (eess.SY).
Pagina del lavoro: https://arxiv.org/abs/2609.27312 — 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 2026-10-08T08:13:26.649Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-24?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).