MA-ROESL: Motion-aware Rapid Reward Optimization for Efficient Robot Skill Learning from Single Videos

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wang, Xianghui, Zhang, Xinming, Chen, Yanjun, Shen, Xiaoyu, Zhang, Wei
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915286330900480
author Wang, Xianghui
Zhang, Xinming
Chen, Yanjun
Shen, Xiaoyu
Zhang, Wei
author_facet Wang, Xianghui
Zhang, Xinming
Chen, Yanjun
Shen, Xiaoyu
Zhang, Wei
contents Vision-language models (VLMs) have demonstrated excellent high-level planning capabilities, enabling locomotion skill learning from video demonstrations without the need for meticulous human-level reward design. However, the improper frame sampling method and low training efficiency of current methods remain a critical bottleneck, resulting in substantial computational overhead and time costs. To address this limitation, we propose Motion-aware Rapid Reward Optimization for Efficient Robot Skill Learning from Single Videos (MA-ROESL). MA-ROESL integrates a motion-aware frame selection method to implicitly enhance the quality of VLM-generated reward functions. It further employs a hybrid three-phase training pipeline that improves training efficiency via rapid reward optimization and derives the final policy through online fine-tuning. Experimental results demonstrate that MA-ROESL significantly enhances training efficiency while faithfully reproducing locomotion skills in both simulated and real-world settings, thereby underscoring its potential as a robust and scalable framework for efficient robot locomotion skill learning from video demonstrations.
format Preprint
id arxiv_https___arxiv_org_abs_2505_08367
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle MA-ROESL: Motion-aware Rapid Reward Optimization for Efficient Robot Skill Learning from Single Videos
Wang, Xianghui
Zhang, Xinming
Chen, Yanjun
Shen, Xiaoyu
Zhang, Wei
Robotics
Vision-language models (VLMs) have demonstrated excellent high-level planning capabilities, enabling locomotion skill learning from video demonstrations without the need for meticulous human-level reward design. However, the improper frame sampling method and low training efficiency of current methods remain a critical bottleneck, resulting in substantial computational overhead and time costs. To address this limitation, we propose Motion-aware Rapid Reward Optimization for Efficient Robot Skill Learning from Single Videos (MA-ROESL). MA-ROESL integrates a motion-aware frame selection method to implicitly enhance the quality of VLM-generated reward functions. It further employs a hybrid three-phase training pipeline that improves training efficiency via rapid reward optimization and derives the final policy through online fine-tuning. Experimental results demonstrate that MA-ROESL significantly enhances training efficiency while faithfully reproducing locomotion skills in both simulated and real-world settings, thereby underscoring its potential as a robust and scalable framework for efficient robot locomotion skill learning from video demonstrations.
title MA-ROESL: Motion-aware Rapid Reward Optimization for Efficient Robot Skill Learning from Single Videos
topic Robotics
url https://arxiv.org/abs/2505.08367