← Back to archive day

Mask World Model: Predicting What Matters for Robust Robot Policy Learning

Yunfan Lou, Xiaowei Chi, Xiaojie Zhang, Zezhong Qian, Chengxuan Li, Rongyu Zhang, Yaoxu Lyu, Guoyu Song, Chuyao Fu, Haoxuan Xu, Pengwei Wang, Shanghang Zhang

35

Recommendation Score

breakthrough🔴 AdvancedReinforcement LearningWorld ModelsBenchmarkUseful for both

Research context

Primary field

Reinforcement Learning

Sequential decision-making, world models, and policy learning.

Topics

World Models

Paper type

Benchmark

Best for

Useful for both

arXiv categories

cs.ROcs.RO

Why It Matters

Mask World Model filters irrelevant visual noise in robot learning, enabling robust policy training from noisy video data. This drastically improves generalization in dynamic real-world environments.

Abstract

World models derived from large-scale video generative pre-training have emerged as a promising paradigm for generalist robot policy learning. However, standard approaches often focus on high-fidelity RGB video prediction, this can result in overfitting to irrelevant factors, such as dynamic backgrounds and illumination changes. These distractions reduce the model's ability to generalize, ultimately leading to unreliable and fragile control policies. To address this, we introduce the Mask World Model (MWM), which leverages video diffusion architectures to predict the evolution of semantic masks instead of pixels. This shift imposes a geometric information bottleneck, forcing the model to capture essential physical dynamics and contact relations while filtering out visual noise. We seamlessly integrate this mask dynamics backbone with a diffusion-based policy head to enable robust end-to-end control. Extensive evaluations demonstrate the superiority of MWM on the LIBERO and RLBench simulation benchmarks, significantly outperforming the state-of-the-art RGB-based world models. Furthermore, real-world experiments and robustness evaluation (via random token pruning) reveal that MWM exhibits superior generalization capabilities and robust resilience to texture information loss.

Published April 21, 2026
© 2026 A2A.pub — AI to Action. From papers to practice, daily.
Summaries are AI-assistedPrivacyTerms