Metraipolis › archive › dc_2ab786a2d1ac

RoboICL: Embodied In-Context Learning with GPT-6 Astra — arXiv:2609.34261

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.34261
License (as archived)
CC0-1.0
Archived
2026-10-07T22:40:25.984Z
SHA-256 of the archived text
254e9ae6195c307fd4b1d521ce8693da078a6c960d7bbfe218bc8d18231a5a7d
Stable URL of this page
https://metraipolis.com/documento/dc_2ab786a2d1ac

How to cite

RoboICL: Embodied In-Context Learning with GPT-6 Astra — arXiv:2609.34261 — arXiv (en), https://arxiv.org/abs/2609.34261 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_2ab786a2d1ac (Metraipolis archive dc_2ab786a2d1ac, sha256 254e9ae6195c307fd4b1d521ce8693da078a6c960d7bbfe218bc8d18231a5a7d)
BibTeX
@misc{dc2ab786a2d1ac,
  title = {RoboICL: Embodied In-Context Learning with GPT-6 Astra — arXiv:2609.34261},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.34261},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_2ab786a2d1ac; sha256 254e9ae6195c307fd4b1d521ce8693da078a6c960d7bbfe218bc8d18231a5a7d},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.34261. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

General-purpose vision-language models offer a promising way to zero-shot robot control: \gptastra{} excels at open-ended and language- or image-conditioned manipulation but remains substantially weaker on high-precision and long-horizon tasks. We introduce \emph{RoboICL}, an in-context robot-control framework that narrows these gaps without robot-specific parameter updates or a learned VLA. RoboICL separates \emph{demonstration context}, which provides recorded examples when available, from \emph{interaction memory}, which accumulates the model's own actions and observed outcomes. Both use a shared observation--action--receipt--observation grammar. To preserve experience across task stages, RoboICL combines sampled demonstration blocks with bounded anchored memory. Fixed anchors keep earlier rollout interactions available for in-context learning, while the latest interaction supports immediate error correction. Across 30 RoboDojo tasks, using zero shot for Open and one demonstration elsewhere, RoboICL improves on official zero-shot \gptastra{} by 20--27 progress-score points in every category. It leads the leaderboard baselines on Memory and Open, achieves comparable performance to the strongest Precision baseline, and remains competitive on Long-Horizon. Its 30-task Overall score is 50.64, versus 33.68 for the strongest baseline. On a separate ten-task subset, RoboICL scores 60.60, within 2.00 points of the $\pi_{0.5}$ + \gptastra{} hybrid approach. On three real-robot tasks, mean progress rises from 14.45 at zero shot to 63.33 at one shot and 78.89 at three shots. On two development tasks, optional Jev-gated action reuse reduces \gptastra{} calls by 33--48\%. Code is available at \href{ this https URL }{ this https URL }.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.LG).
Categorie arXiv: Robotics (cs.RO) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.34261 — 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 2026-10-07T22:14:28.824Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).