Metraipolis › archive › dc_ccbdfb97cdcd

JEDI: Joint-Embedding Demonstration Compression for In-Context Imitation Learning — arXiv:2610.40356

Source: arXiv (en)
Original: https://arxiv.org/abs/2610.40356
License (as archived): CC0-1.0
Archived: 2026-11-01T02:45:48.257Z
SHA-256 of the archived text: 5701c1a422e0109083a0209b6f84d913d3607f4624224af4efbb9d01b033aeba
Stable URL of this page: https://metraipolis.com/documento/dc_ccbdfb97cdcd

How to cite

JEDI: Joint-Embedding Demonstration Compression for In-Context Imitation Learning — arXiv:2610.40356 — arXiv (en), https://arxiv.org/abs/2610.40356 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ccbdfb97cdcd (Metraipolis archive dc_ccbdfb97cdcd, sha256 5701c1a422e0109083a0209b6f84d913d3607f4624224af4efbb9d01b033aeba)

BibTeX

@misc{dcccbdfb97cdcd,
  title = {JEDI: Joint-Embedding Demonstration Compression for In-Context Imitation Learning — arXiv:2610.40356},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.40356},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ccbdfb97cdcd; sha256 5701c1a422e0109083a0209b6f84d913d3607f4624224af4efbb9d01b033aeba},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.40356. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

In-context imitation learning (ICIL) adapts robot policies through test-time demonstrations, but long multimodal trajectories constrain how many examples fit in context. We ask what a compressed demonstration should preserve. We propose JEDI, which compresses demonstrations into latent slots trained to predict the query episode's next-stage keyframe in a frozen video-JEPA space. Task and process counterfactuals help the slots preserve execution details, while the predictor selects demonstrations from a bank. Post-training with keypoint tokens enables human-hand conditioning on robots. Experiments on LIBERO and real robots show that JEDI supports many-shot context; LIBERO objective controls show improved success and demonstration following over action-only compression. These findings support query-future prediction as a training objective for preserving execution-relevant information in compressed demonstrations.

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/2610.40356 — 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 2026-11-01T02:16:58.683Z dalla pagina https://arxiv.org/list/cs.RO/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).