Adversarial Fine-tuning in Offline-to-Online Reinforcement Learning for Robust Robot Control

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ayabe, Shingo, Kera, Hiroshi, Kawamoto, Kazuhiko
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911473525063680
author Ayabe, Shingo
Kera, Hiroshi
Kawamoto, Kazuhiko
author_facet Ayabe, Shingo
Kera, Hiroshi
Kawamoto, Kazuhiko
contents Offline reinforcement learning enables sample-efficient policy acquisition without risky online interaction, yet policies trained on static datasets remain brittle under action-space perturbations such as actuator faults. This study introduces an offline-to-online framework that trains policies on clean data and then performs adversarial fine-tuning, where perturbations are injected into executed actions to induce compensatory behavior and improve resilience. A performance-aware curriculum further adjusts the perturbation probability during training via an exponential-moving-average signal, balancing robustness and stability throughout the learning process. Experiments on continuous-control locomotion tasks demonstrate that the proposed method consistently improves robustness over offline-only baselines and converges faster than training from scratch. Matching the fine-tuning and evaluation conditions yields the strongest robustness to action-space perturbations, while the adaptive curriculum strategy mitigates the degradation of nominal performance observed with the linear curriculum strategy. Overall, the results show that adversarial fine-tuning enables adaptive and robust control under uncertain environments, bridging the gap between offline efficiency and online adaptability.
format Preprint
id arxiv_https___arxiv_org_abs_2510_13358
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Adversarial Fine-tuning in Offline-to-Online Reinforcement Learning for Robust Robot Control
Ayabe, Shingo
Kera, Hiroshi
Kawamoto, Kazuhiko
Robotics
Artificial Intelligence
Offline reinforcement learning enables sample-efficient policy acquisition without risky online interaction, yet policies trained on static datasets remain brittle under action-space perturbations such as actuator faults. This study introduces an offline-to-online framework that trains policies on clean data and then performs adversarial fine-tuning, where perturbations are injected into executed actions to induce compensatory behavior and improve resilience. A performance-aware curriculum further adjusts the perturbation probability during training via an exponential-moving-average signal, balancing robustness and stability throughout the learning process. Experiments on continuous-control locomotion tasks demonstrate that the proposed method consistently improves robustness over offline-only baselines and converges faster than training from scratch. Matching the fine-tuning and evaluation conditions yields the strongest robustness to action-space perturbations, while the adaptive curriculum strategy mitigates the degradation of nominal performance observed with the linear curriculum strategy. Overall, the results show that adversarial fine-tuning enables adaptive and robust control under uncertain environments, bridging the gap between offline efficiency and online adaptability.
title Adversarial Fine-tuning in Offline-to-Online Reinforcement Learning for Robust Robot Control
topic Robotics
Artificial Intelligence
url https://arxiv.org/abs/2510.13358