Metraipolis › archive › dc_df84720cffd9

Policy-Calibrated DAgger: Offline Calibrated Noise Injection for Imitation Learning — arXiv:2609.30462

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.30462
License (as archived)
CC0-1.0
Archived
2026-10-08T03:14:40.591Z
SHA-256 of the archived text
965d56ec27e10afe7a5519b6da17848def1ea7f3a5e5e817d569306d9d2acc27
Stable URL of this page
https://metraipolis.com/documento/dc_df84720cffd9

How to cite

Policy-Calibrated DAgger: Offline Calibrated Noise Injection for Imitation Learning — arXiv:2609.30462 — arXiv (en), https://arxiv.org/abs/2609.30462 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_df84720cffd9 (Metraipolis archive dc_df84720cffd9, sha256 965d56ec27e10afe7a5519b6da17848def1ea7f3a5e5e817d569306d9d2acc27)
BibTeX
@misc{dcdf84720cffd9,
  title = {Policy-Calibrated DAgger: Offline Calibrated Noise Injection for Imitation Learning — arXiv:2609.30462},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.30462},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_df84720cffd9; sha256 965d56ec27e10afe7a5519b6da17848def1ea7f3a5e5e817d569306d9d2acc27},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.30462. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Policies trained with imitation learning can accumulate errors over time, causing the robot to drift outside the training distribution. Existing methods mitigate this covariate shift by collecting additional data where the policy fails or is likely to fail. The first places the robot in unsafe conditions and the second requires choosing an appropriate noise distribution to collect new expert demonstrations under that noise. We propose Policy-Calibrated DAgger, a method that makes use of the properties of recent generative policies to estimate the policy's noise offline by using its own predicted action distribution. We measure a diffusion policy's spread of predicted actions at observations along the expert trajectory and measure its closed-loop error relative to a recorded trajectory. To address issues with measuring error in a multimodal action space, we guide the policy towards the trajectory during closed-loop control through partial denoising, and use properties of a diffusion model to unnormalize the measured error as if we did not guide it. We experiment in a scenario where a robot is tasked to reach an engine lever in a cluttered and narrow environment and show results in a 3D photorealistic simulator and a 2D planar reacher environment. We show that our method surpasses policies trained with dataset aggregation without noising and matches the performance of the best noise level in hindsight, without requiring a sweep over noise levels.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-28 in cs.RO).
Categorie arXiv: Robotics (cs.RO).
Pagina del lavoro: https://arxiv.org/abs/2609.30462 — 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 2026-10-08T02:58:08.786Z dalla pagina https://arxiv.org/catchup/cs.RO/2026-09-28?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).