Keep on Going: Learning Robust Humanoid Motion Skills via Selective Adversarial Training

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhang, Yang, Cao, Zhanxiang, Nie, Buqing, Li, Haoyang, Jiangwei, Zhong, Sun, Qiao, Hu, Xiaoyi, Yang, Xiaokang, Gao, Yue
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908649613426688
author Zhang, Yang
Cao, Zhanxiang
Nie, Buqing
Li, Haoyang
Jiangwei, Zhong
Sun, Qiao
Hu, Xiaoyi
Yang, Xiaokang
Gao, Yue
author_facet Zhang, Yang
Cao, Zhanxiang
Nie, Buqing
Li, Haoyang
Jiangwei, Zhong
Sun, Qiao
Hu, Xiaoyi
Yang, Xiaokang
Gao, Yue
contents Humanoid robots are expected to operate reliably over long horizons while executing versatile whole-body skills. Yet Reinforcement Learning (RL) motion policies typically lose stability under prolonged operation, sensor/actuator noise, and real world disturbances. In this work, we propose a Selective Adversarial Attack for Robust Training (SA2RT) to enhance the robustness of motion skills. The adversary is learned to identify and sparsely perturb the most vulnerable states and actions under an attack-budget constraint, thereby exposing true weakness without inducing conservative overfitting. The resulting non-zero sum, alternating optimization continually strengthens the motion policy against the strongest discovered attacks. We validate our approach on the Unitree G1 humanoid robot across perceptive locomotion and whole-body control tasks. Experimental results show that adversarially trained policies improve the terrain traversal success rate by 40%, reduce the trajectory tracking error by 32%, and maintain long horizon mobility and tracking performance. Together, these results demonstrate that selective adversarial attacks are an effective driver for learning robust, long horizon humanoid motion skills.
format Preprint
id arxiv_https___arxiv_org_abs_2507_08303
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Keep on Going: Learning Robust Humanoid Motion Skills via Selective Adversarial Training
Zhang, Yang
Cao, Zhanxiang
Nie, Buqing
Li, Haoyang
Jiangwei, Zhong
Sun, Qiao
Hu, Xiaoyi
Yang, Xiaokang
Gao, Yue
Robotics
Humanoid robots are expected to operate reliably over long horizons while executing versatile whole-body skills. Yet Reinforcement Learning (RL) motion policies typically lose stability under prolonged operation, sensor/actuator noise, and real world disturbances. In this work, we propose a Selective Adversarial Attack for Robust Training (SA2RT) to enhance the robustness of motion skills. The adversary is learned to identify and sparsely perturb the most vulnerable states and actions under an attack-budget constraint, thereby exposing true weakness without inducing conservative overfitting. The resulting non-zero sum, alternating optimization continually strengthens the motion policy against the strongest discovered attacks. We validate our approach on the Unitree G1 humanoid robot across perceptive locomotion and whole-body control tasks. Experimental results show that adversarially trained policies improve the terrain traversal success rate by 40%, reduce the trajectory tracking error by 32%, and maintain long horizon mobility and tracking performance. Together, these results demonstrate that selective adversarial attacks are an effective driver for learning robust, long horizon humanoid motion skills.
title Keep on Going: Learning Robust Humanoid Motion Skills via Selective Adversarial Training
topic Robotics
url https://arxiv.org/abs/2507.08303