Metraipolis › archive › dc_25f4697242e2
Why Koopman Structure Does Not Help Differentiable World Models for Humanoid Whole-Body Tracking — arXiv:2611.00951
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.00951
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-03T11:44:24.301Z
- SHA-256 of the archived text
2ac504694899524905e0c19daa0b091995ed21c04eeca156417dcd78b8b4a721- Stable URL of this page
https://metraipolis.com/documento/dc_25f4697242e2
How to cite
Why Koopman Structure Does Not Help Differentiable World Models for Humanoid Whole-Body Tracking — arXiv:2611.00951 — arXiv (en), https://arxiv.org/abs/2611.00951 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_25f4697242e2 (Metraipolis archive dc_25f4697242e2, sha256 2ac504694899524905e0c19daa0b091995ed21c04eeca156417dcd78b8b4a721)
BibTeX
@misc{dc25f4697242e2,
title = {Why Koopman Structure Does Not Help Differentiable World Models for Humanoid Whole-Body Tracking — arXiv:2611.00951},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.00951},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_25f4697242e2; sha256 2ac504694899524905e0c19daa0b091995ed21c04eeca156417dcd78b8b4a721},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.00951. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
We test whether Koopman structure, linear dynamics in a lifted state space, helps when a humanoid whole-body tracking policy is trained by backpropagating through a differentiable world model, replacing the MLP world model of an otherwise unchanged pipeline for simulated Unitree G1 tracking. It does not: return is at best equal (0.823 versus 0.830), one-step error on identical data is 1.3 to 1.6 times the MLP's, and training is about 15% slower. Lifted rollouts diverge or fail to train a policy; of the schemes tested, re-lifting at every imagined step trains best. Re-lifting never uses the lifted linear dynamics and leaves a model affine in the action, with state-dependent offset and slope; an action-affine MLP also predicts worse, so the cost belongs to this class. At a fixed state, the curvature of the true dynamics in the action explains only part of it. Across states, the policy's actions span a curved region, so an affine model fits policy data only as the tangent at the policy's action, the operating point: on G1, the MLP's tangent there is as accurate as the MLP; at the mean action, 31 times worse. The offset must encode, from the state alone, where the policy acts, whereas an MLP reads the action. Supplying the policy's action removes most of the remaining gap on G1, Ant and Humanoid. The only online repair approaching the MLP feeds the action into the nonlinear lift, giving up the structure. The MLP world model thus remains the better choice here. 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/2611.00951 — 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 2026-11-03T11:06:52.248Z dalla pagina https://arxiv.org/list/cs.RO/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).