Metraipolis › archive › dc_ea87188f25e8

Principal Steering Subspaces for Online Adaptation of Frozen Generative Robot Policies — arXiv:2609.33765

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.33765
License (as archived)
CC0-1.0
Archived
2026-10-08T00:17:13.621Z
SHA-256 of the archived text
61149ec3820a47078fb1b0a12df83a0b43cb8c64d30bd68ca0c26beb20d5df7a
Stable URL of this page
https://metraipolis.com/documento/dc_ea87188f25e8

How to cite

Principal Steering Subspaces for Online Adaptation of Frozen Generative Robot Policies — arXiv:2609.33765 — arXiv (en), https://arxiv.org/abs/2609.33765 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ea87188f25e8 (Metraipolis archive dc_ea87188f25e8, sha256 61149ec3820a47078fb1b0a12df83a0b43cb8c64d30bd68ca0c26beb20d5df7a)
BibTeX
@misc{dcea87188f25e8,
  title = {Principal Steering Subspaces for Online Adaptation of Frozen Generative Robot Policies — arXiv:2609.33765},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.33765},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ea87188f25e8; sha256 61149ec3820a47078fb1b0a12df83a0b43cb8c64d30bd68ca0c26beb20d5df7a},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.33765. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Generative robot policies provide expressive behavior priors, but updating a large diffusion or flow-matching model through online interaction is costly. Latent-space reinforcement learning avoids updating the pretrained generator by controlling its initial sampling noise, yet high-dimensional noise can have strongly anisotropic effects on decoded actions. We introduce Principal Steering Subspaces (PSS), a forward-query interface that constructs a fixed low-dimensional control basis from finite-difference decoder responses. Soft Actor-Critic controls the leading response directions, while the orthogonal complement is independently resampled from the Gaussian prior at each query. On three RoboMimic tasks with diffusion and flow-matching policies, response spectra reveal substantial concentration. Across five matched task-generator pairs, the training curves indicate that PSS generally converges faster and exhibits more stable late-training behavior than full-latent control, while achieving stronger final performance overall. Controlled Diffusion-Square ablations further show that leading-response directions outperform random and least-responsive subspaces of equal dimension. We further integrate PSS with a frozen, closed-source 3B-parameter vision-language-action (VLA) policy in a humanoid learning system with synchronous transition collection, reset-time optimization, and latency-aware asynchronous deployment. In an exploratory screwdriver-placement evaluation, success is observed in 2/10 trials for the frozen VLA policy and 6/10 after SAC+PSS adaptation. These results support decoder-response geometry as a practical basis for online adaptation of frozen generative robot policies.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-29 in cs.RO).
Categorie arXiv: Robotics (cs.RO).
Pagina del lavoro: https://arxiv.org/abs/2609.33765 — 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 2026-10-07T23:52:16.496Z dalla pagina https://arxiv.org/catchup/cs.RO/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).