Metraipolis › archive › dc_2167c75b0fea
FOCUS: First-Order Control Using Optimal Transport Supervision for Imitation from Observations — arXiv:2610.41540
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.41540
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-02T07:50:53.068Z
- SHA-256 of the archived text
878f4bf4bdcf6e3cff87fca1e0267f944215136d25f9797030e49a14b4602d01- Stable URL of this page
https://metraipolis.com/documento/dc_2167c75b0fea
How to cite
FOCUS: First-Order Control Using Optimal Transport Supervision for Imitation from Observations — arXiv:2610.41540 — arXiv (en), https://arxiv.org/abs/2610.41540 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_2167c75b0fea (Metraipolis archive dc_2167c75b0fea, sha256 878f4bf4bdcf6e3cff87fca1e0267f944215136d25f9797030e49a14b4602d01)
BibTeX
@misc{dc2167c75b0fea,
title = {FOCUS: First-Order Control Using Optimal Transport Supervision for Imitation from Observations — arXiv:2610.41540},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.41540},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_2167c75b0fea; sha256 878f4bf4bdcf6e3cff87fca1e0267f944215136d25f9797030e49a14b4602d01},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.41540. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Imitation learning from observations must recover behavior from state-only demonstrations without known temporal correspondence. Differentiable simulators provide pathwise policy gradients, but differentiating through long, contact-rich trajectories can make optimization unstable. We introduce FOCUS, which combines local optimal-transport alignment with short-horizon differentiable actor-critic learning. FOCUS matches agent rollout windows to multiple sampled expert segments using a soft best-of-$K$ objective that emphasizes locally compatible matches. The resulting transport-based costs provide differentiable imitation rewards, while a learned critic bootstraps value beyond the rollout horizon. FOCUS requires neither temporally aligned trajectories nor expert-aligned initialization or teacher forcing. Across contact-rich locomotion tasks in DFlex, including a musculoskeletal humanoid controlled by 152 muscle activations, FOCUS improves sample efficiency and training stability over adversarial imitation-from-observation methods and prior differentiable imitation baselines. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG) ; Robotics (cs.RO). Pagina del lavoro: https://arxiv.org/abs/2610.41540 — 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 2026-11-02T07:40:00.197Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).