An Real-Sim-Real (RSR) Loop Framework for Generalizable Robotic Policy Transfer with Differentiable Simulation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Shi, Lu, Xu, Yuxuan, Wang, Shiyu, Huang, Jinhao, Zhao, Wenhao, Jia, Yufei, Yan, Zike, Gu, Weibin, Zhou, Guyue
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910879923044352
author Shi, Lu
Xu, Yuxuan
Wang, Shiyu
Huang, Jinhao
Zhao, Wenhao
Jia, Yufei
Yan, Zike
Gu, Weibin
Zhou, Guyue
author_facet Shi, Lu
Xu, Yuxuan
Wang, Shiyu
Huang, Jinhao
Zhao, Wenhao
Jia, Yufei
Yan, Zike
Gu, Weibin
Zhou, Guyue
contents The sim-to-real gap remains a critical challenge in robotics, hindering the deployment of algorithms trained in simulation to real-world systems. This paper introduces a novel Real-Sim-Real (RSR) loop framework leveraging differentiable simulation to address this gap by iteratively refining simulation parameters, aligning them with real-world conditions, and enabling robust and efficient policy transfer. A key contribution of our work is the design of an informative cost function that encourages the collection of diverse and representative real-world data, minimizing bias and maximizing the utility of each data point for simulation refinement. This cost function integrates seamlessly into existing reinforcement learning algorithms (e.g., PPO, SAC) and ensures a balanced exploration of critical regions in the real domain. Furthermore, our approach is implemented on the versatile Mujoco MJX platform, and our framework is compatible with a wide range of robotic systems. Experimental results on several robotic manipulation tasks demonstrate that our method significantly reduces the sim-to-real gap, achieving high task performance and generalizability across diverse scenarios of both explicit and implicit environmental uncertainties.
format Preprint
id arxiv_https___arxiv_org_abs_2503_10118
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle An Real-Sim-Real (RSR) Loop Framework for Generalizable Robotic Policy Transfer with Differentiable Simulation
Shi, Lu
Xu, Yuxuan
Wang, Shiyu
Huang, Jinhao
Zhao, Wenhao
Jia, Yufei
Yan, Zike
Gu, Weibin
Zhou, Guyue
Robotics
Machine Learning
The sim-to-real gap remains a critical challenge in robotics, hindering the deployment of algorithms trained in simulation to real-world systems. This paper introduces a novel Real-Sim-Real (RSR) loop framework leveraging differentiable simulation to address this gap by iteratively refining simulation parameters, aligning them with real-world conditions, and enabling robust and efficient policy transfer. A key contribution of our work is the design of an informative cost function that encourages the collection of diverse and representative real-world data, minimizing bias and maximizing the utility of each data point for simulation refinement. This cost function integrates seamlessly into existing reinforcement learning algorithms (e.g., PPO, SAC) and ensures a balanced exploration of critical regions in the real domain. Furthermore, our approach is implemented on the versatile Mujoco MJX platform, and our framework is compatible with a wide range of robotic systems. Experimental results on several robotic manipulation tasks demonstrate that our method significantly reduces the sim-to-real gap, achieving high task performance and generalizability across diverse scenarios of both explicit and implicit environmental uncertainties.
title An Real-Sim-Real (RSR) Loop Framework for Generalizable Robotic Policy Transfer with Differentiable Simulation
topic Robotics
Machine Learning
url https://arxiv.org/abs/2503.10118