Metraipolis › archive › dc_9db255ceaf7d
FOCUS: From Privileged States to RGB-D with Controlled Modality Switching and Representation Alignment — arXiv:2610.11119
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.11119
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-09T06:33:13.000Z
- SHA-256 of the archived text
3da232fe15d9e81f2726f5dd6fb372d5fd8d468cef6c979a4b32d591be83b838- Stable URL of this page
https://metraipolis.com/documento/dc_9db255ceaf7d
How to cite
FOCUS: From Privileged States to RGB-D with Controlled Modality Switching and Representation Alignment — arXiv:2610.11119 — arXiv (en), https://arxiv.org/abs/2610.11119 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_9db255ceaf7d (Metraipolis archive dc_9db255ceaf7d, sha256 3da232fe15d9e81f2726f5dd6fb372d5fd8d468cef6c979a4b32d591be83b838)
BibTeX
@misc{dc9db255ceaf7d,
title = {FOCUS: From Privileged States to RGB-D with Controlled Modality Switching and Representation Alignment — arXiv:2610.11119},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.11119},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_9db255ceaf7d; sha256 3da232fe15d9e81f2726f5dd6fb372d5fd8d468cef6c979a4b32d591be83b838},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.11119. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Vision-based reinforcement learning for robotic manipulation is sample-inefficient because RGB-D observations are high-dimensional and noisy. Privileged state information available in simulation can accelerate training, but its absence at test time creates a train-test modality gap. We propose FOCUS, a single-stage PPO framework that trains the critic on privileged state while automatically regulating whether the actor collects rollouts from RGB-D or privileged-state latents. Regulation is driven by the KL divergence between the action distributions induced by the two modalities, while representation alignment encourages consistent action selection across them. Together, these mechanisms limit RGB-D rollouts when the actor's action distributions from RGB-D and privileged-state latents disagree. As they align, RGB-D exposure increases, shifting on-policy training toward the RGB-D inputs used at test time. Across five manipulation tasks, FOCUS raises average test success from 0.71 to 0.93 relative to the strongest RGB-D-at-test baseline on each task. When accounting for each method's complete training pipeline, budget-normalized training-success AUC increases from 0.47 to 0.65. On Pick-and-Place, test success rises from 0.47 to 0.86, while AUC increases from 0.12 to 0.61, a 5.0x improvement in learning efficiency over the fixed interaction budget. Tema degli esploratori: statistica e scienza dei dati (elenco arXiv cs.RO, nuovi lavori del giorno). Categorie arXiv: Robotics (cs.RO). Pagina del lavoro: https://arxiv.org/abs/2610.11119 — 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 2026-10-09T05:55:22.002Z dalla pagina https://arxiv.org/list/cs.RO/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).