Can World and Action Modeling Benefit Each Other under Unified Causal Training?

This study explores whether world and action modeling can mutually benefit each other under unified causal training. A simple framework, OneWM, is introduced that treats actions as both conditions for world prediction and targets for policy learning within a shared video-action architecture. The research demonstrates that learning the visual consequences of specified actions improves policy performance and generalization across changing scenes, while action learning enhances multi-view world-prediction fidelity under the same architecture.