Metraipolis › archive › dc_abc47a0229b2

JEPA Policy: Diffusion-Free Imitation Learning via Paired Action and Future Representation Prediction — arXiv:2609.09630

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.09630
License (as archived)
CC0-1.0
Archived
2026-10-16T11:11:35.143Z
SHA-256 of the archived text
0e5e021a8b85f6924655644f14581094c9844ca7d3166fbfed4434d62b6bb3a9
Stable URL of this page
https://metraipolis.com/documento/dc_abc47a0229b2

How to cite

JEPA Policy: Diffusion-Free Imitation Learning via Paired Action and Future Representation Prediction — arXiv:2609.09630 — arXiv (en), https://arxiv.org/abs/2609.09630 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_abc47a0229b2 (Metraipolis archive dc_abc47a0229b2, sha256 0e5e021a8b85f6924655644f14581094c9844ca7d3166fbfed4434d62b6bb3a9)
BibTeX
@misc{dcabc47a0229b2,
  title = {JEPA Policy: Diffusion-Free Imitation Learning via Paired Action and Future Representation Prediction — arXiv:2609.09630},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.09630},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_abc47a0229b2; sha256 0e5e021a8b85f6924655644f14581094c9844ca7d3166fbfed4434d62b6bb3a9},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.09630. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Standard behavior cloning supervises actions without explicitly constraining the future representation paired with each demonstrated action chunk. We introduce JEPA Policy, a diffusion-free framework that uses the action chunk and its observed future representation as paired training targets. Action and future-representation tokens interact in a shared Transformer and are refined through two forward passes. Future prediction can therefore shape the representation used to generate actions. Dual-branch and gradient-routing controls attribute the gain to this shared topology rather than to an auxiliary prediction head alone. Across nine simulated tasks, JEPA Policy improves mean success over the action-only MIP baseline and outperforms Diffusion Policy under the evaluated configurations, while adding 0.29 ms to MIP's model latency. A five-task, 630-episode physical-robot study produces the same pooled ranking. Further audits find no complete representation collapse under action supervision and identify a task-conditioned failure-ranking signal in future-prediction error. These results support paired future-representation supervision as a practical approach to low-latency visuomotor imitation without iterative generative sampling.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-10 in cs.RO).
Categorie arXiv: Robotics (cs.RO).
Pagina del lavoro: https://arxiv.org/abs/2609.09630 — 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 2026-10-16T10:19:11.885Z dalla pagina https://arxiv.org/catchup/cs.RO/2026-09-10?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).