Salvato in:
Dettagli Bibliografici
Autori principali: Tong, Dezhong, Wang, Jiawen, Zhou, Hengyi, Shen, Yinglong, Huang, Xiaonan, Jawed, M. Khalid
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:https://arxiv.org/abs/2605.03288
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911646706827264
author Tong, Dezhong
Wang, Jiawen
Zhou, Hengyi
Shen, Yinglong
Huang, Xiaonan
Jawed, M. Khalid
author_facet Tong, Dezhong
Wang, Jiawen
Zhou, Hengyi
Shen, Yinglong
Huang, Xiaonan
Jawed, M. Khalid
contents Many physical AI tasks are governed by implicit equilibrium: an agent actuates a subset of degrees of freedom (boundary DoFs), while the remaining free DoFs settle by minimizing a total potential energy. Even seemingly basic tasks such as bending a deformable linear object (DLO) to a target shape can exhibit strongly nonlinear behavior due to multi-stability: the same boundary conditions may yield multiple equilibrium shapes depending on the actuation trajectory. However, learning and control in such systems is brittle because the actuation-to-configuration map is defined only implicitly, and naive backpropagation through iterative equilibrium solvers is memory- and compute-intensive. We propose Neural Control, a boundary-control framework that computes trajectory-dependent, memory-efficient proxy gradients by differentiating equilibrium conditions via an adjoint formulation, avoiding unrolling of solver iterations. To improve robustness over long horizons, we integrate these sensitivities into a receding-horizon MPC scheme that repeatedly re-anchors optimization to realized equilibria and mitigates basin-switching in multi-stable regimes. We evaluate Neural Control in simulation and on physical robots manipulating DLOs, and show improved performance over gradient-free baselines such as SPSA and CEM.
format Preprint
id arxiv_https___arxiv_org_abs_2605_03288
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Neural Control: Adjoint Learning Through Equilibrium Constraints
Tong, Dezhong
Wang, Jiawen
Zhou, Hengyi
Shen, Yinglong
Huang, Xiaonan
Jawed, M. Khalid
Robotics
Many physical AI tasks are governed by implicit equilibrium: an agent actuates a subset of degrees of freedom (boundary DoFs), while the remaining free DoFs settle by minimizing a total potential energy. Even seemingly basic tasks such as bending a deformable linear object (DLO) to a target shape can exhibit strongly nonlinear behavior due to multi-stability: the same boundary conditions may yield multiple equilibrium shapes depending on the actuation trajectory. However, learning and control in such systems is brittle because the actuation-to-configuration map is defined only implicitly, and naive backpropagation through iterative equilibrium solvers is memory- and compute-intensive. We propose Neural Control, a boundary-control framework that computes trajectory-dependent, memory-efficient proxy gradients by differentiating equilibrium conditions via an adjoint formulation, avoiding unrolling of solver iterations. To improve robustness over long horizons, we integrate these sensitivities into a receding-horizon MPC scheme that repeatedly re-anchors optimization to realized equilibria and mitigates basin-switching in multi-stable regimes. We evaluate Neural Control in simulation and on physical robots manipulating DLOs, and show improved performance over gradient-free baselines such as SPSA and CEM.
title Neural Control: Adjoint Learning Through Equilibrium Constraints
topic Robotics
url https://arxiv.org/abs/2605.03288