Metraipolis › archive › dc_bf7b9acbe032

De-Diffusing WAMs: Action-Sufficient Direct Prediction for Robot Control — arXiv:2610.28122

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.28122
License (as archived)
CC0-1.0
Archived
2026-10-21T07:08:15.628Z
SHA-256 of the archived text
4ae17440fc9e36e0bd7a6dcbee035de7ec45d3aaaf570416dac2922aedae0354
Stable URL of this page
https://metraipolis.com/documento/dc_bf7b9acbe032

How to cite

De-Diffusing WAMs: Action-Sufficient Direct Prediction for Robot Control — arXiv:2610.28122 — arXiv (en), https://arxiv.org/abs/2610.28122 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_bf7b9acbe032 (Metraipolis archive dc_bf7b9acbe032, sha256 4ae17440fc9e36e0bd7a6dcbee035de7ec45d3aaaf570416dac2922aedae0354)
BibTeX
@misc{dcbf7b9acbe032,
  title = {De-Diffusing WAMs: Action-Sufficient Direct Prediction for Robot Control — arXiv:2610.28122},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.28122},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_bf7b9acbe032; sha256 4ae17440fc9e36e0bd7a6dcbee035de7ec45d3aaaf570416dac2922aedae0354},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.28122. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Cascade world-action models (WAMs) use predicted visual futures to condition robot actions, but inherit costly iterative inference from diffusion-pretrained experts. We ask what a single-step predicted future must preserve to support accurate single-step action prediction. In the evaluated WAM, resampling video noise has little effect on downstream actions, yet naively truncating both streams to one step substantially reduces task success. Video-space supervision only partially closes this gap, and a local sensitivity analysis shows that vision prediction errors have highly unequal consequences for action accuracy. These findings motivate DirectWAM, which converts pretrained video and action experts into single-step predictors without stochastic sampling through action-sufficient distillation, while retaining the explicit future-to-action cascade. After learning a one-step action model, we train the future predictor by decoding actions from its own predicted futures and backpropagating demonstration action error through the expert, with an optional future-latent reconstruction loss. The resulting objective evaluates the student-generated future through its downstream control consequences rather than relying on representation matching alone. DirectWAM achieves 97.1% average success across four LIBERO suites and 85.4% on RoboTwin 2.0, outperforming Flash-WAM at the same one-evaluation-per-stream budget. This advantage has also extended to three real-world tasks that we designed. Compared with the full iterative teacher, it reduces inference latency by $28.0\times$.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Robotics (cs.RO) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2610.28122 — 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 2026-10-21T07:05:43.623Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).