Saved in:
Bibliographic Details
Main Authors: Shen, Yutong, Liu, Hangxu, Pei, Kailin, Xia, Ruizhe, Feng, Tongtong
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2601.17507
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915752966094848
author Shen, Yutong
Liu, Hangxu
Pei, Kailin
Xia, Ruizhe
Feng, Tongtong
author_facet Shen, Yutong
Liu, Hangxu
Pei, Kailin
Xia, Ruizhe
Feng, Tongtong
contents Humanoid robot loco-manipulation remains constrained by the semantic-physical gap. Current methods face three limitations: Low sample efficiency in reinforcement learning, poor generalization in imitation learning, and physical inconsistency in VLMs. We propose MetaWorld, a hierarchical world model that integrates semantic planning and physical control via expert policy transfer. The framework decouples tasks into a VLM-driven semantic layer and a latent dynamics model operating in a compact state space. Our dynamic expert selection and motion prior fusion mechanism leverages a pre-trained multi-expert policy library as transferable knowledge, enabling efficient online adaptation via a two-stage framework. VLMs serve as semantic interfaces, mapping instructions to executable skills and bypassing symbol grounding. Experiments on Humanoid-Bench show MetaWorld outperforms world model-based RL in task completion and motion coherence. Our code will be found at https://anonymous.4open.science/r/metaworld-2BF4/
format Preprint
id arxiv_https___arxiv_org_abs_2601_17507
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle MetaWorld: Skill Transfer and Composition in a Hierarchical World Model for Grounding High-Level Instructions
Shen, Yutong
Liu, Hangxu
Pei, Kailin
Xia, Ruizhe
Feng, Tongtong
Robotics
Humanoid robot loco-manipulation remains constrained by the semantic-physical gap. Current methods face three limitations: Low sample efficiency in reinforcement learning, poor generalization in imitation learning, and physical inconsistency in VLMs. We propose MetaWorld, a hierarchical world model that integrates semantic planning and physical control via expert policy transfer. The framework decouples tasks into a VLM-driven semantic layer and a latent dynamics model operating in a compact state space. Our dynamic expert selection and motion prior fusion mechanism leverages a pre-trained multi-expert policy library as transferable knowledge, enabling efficient online adaptation via a two-stage framework. VLMs serve as semantic interfaces, mapping instructions to executable skills and bypassing symbol grounding. Experiments on Humanoid-Bench show MetaWorld outperforms world model-based RL in task completion and motion coherence. Our code will be found at https://anonymous.4open.science/r/metaworld-2BF4/
title MetaWorld: Skill Transfer and Composition in a Hierarchical World Model for Grounding High-Level Instructions
topic Robotics
url https://arxiv.org/abs/2601.17507