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

TL;DR

The paper introduces the Mask World Model (MWM), a novel approach that predicts semantic masks instead of pixels to improve the robustness and generalization of robot policies, outperforming RGB-based models.

Contribution

The paper proposes MWM, which leverages video diffusion architectures to focus on essential physical dynamics, reducing overfitting to irrelevant visual details for better robot control.

Findings

MWM outperforms RGB-based models on LIBERO and RLBench benchmarks.

MWM demonstrates superior robustness to visual noise and texture loss.

Real-world experiments confirm improved generalization and resilience.

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.