Saved in:
Bibliographic Details
Main Authors: Cha, Junhyeok Rui, Cha, Woohyun, Shin, Jaeyong, Kim, Donghyeon, Park, Jaeheung
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.21853
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912982008594432
author Cha, Junhyeok Rui
Cha, Woohyun
Shin, Jaeyong
Kim, Donghyeon
Park, Jaeheung
author_facet Cha, Junhyeok Rui
Cha, Woohyun
Shin, Jaeyong
Kim, Donghyeon
Park, Jaeheung
contents This paper proposes a novel alternative to existing sim-to-real methods for training control policies with simulated experiences. Unlike prior methods that typically rely on domain randomization over a fixed finite set of parameters, the proposed approach injects state-dependent perturbations into the input joint torque during forward simulation. These perturbations are designed to simulate a broader spectrum of reality gaps than standard parameter randomization without requiring additional training. By using neural networks as flexible perturbation generators, the proposed method can represent complex, state-dependent uncertainties, such as nonlinear actuator dynamics and contact compliance, that parametric randomization cannot capture. Experimental results demonstrate that the proposed approach enables humanoid locomotion policies to achieve superior robustness against complex, unseen reality gaps in both simulation and real-world deployment.
format Preprint
id arxiv_https___arxiv_org_abs_2603_21853
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Sim-to-Real of Humanoid Locomotion Policies via Joint Torque Space Perturbation Injection
Cha, Junhyeok Rui
Cha, Woohyun
Shin, Jaeyong
Kim, Donghyeon
Park, Jaeheung
Robotics
Artificial Intelligence
This paper proposes a novel alternative to existing sim-to-real methods for training control policies with simulated experiences. Unlike prior methods that typically rely on domain randomization over a fixed finite set of parameters, the proposed approach injects state-dependent perturbations into the input joint torque during forward simulation. These perturbations are designed to simulate a broader spectrum of reality gaps than standard parameter randomization without requiring additional training. By using neural networks as flexible perturbation generators, the proposed method can represent complex, state-dependent uncertainties, such as nonlinear actuator dynamics and contact compliance, that parametric randomization cannot capture. Experimental results demonstrate that the proposed approach enables humanoid locomotion policies to achieve superior robustness against complex, unseen reality gaps in both simulation and real-world deployment.
title Sim-to-Real of Humanoid Locomotion Policies via Joint Torque Space Perturbation Injection
topic Robotics
Artificial Intelligence
url https://arxiv.org/abs/2603.21853