WMPO: World Model-based Policy Optimization for Vision-Language-Action Models
This work addresses the challenge of sample-efficient learning for general-purpose robotic manipulation, enabling VLA models to learn from failures and self-correct without costly real-world interactions, representing a novel method for a known bottleneck rather than a foundational breakthrough.
The paper tackled the problem of high sample complexity in reinforcement learning for vision-language-action models by introducing World-Model-based Policy Optimization (WMPO), a framework that uses pixel-based world models to enable on-policy learning without real environment interactions, resulting in substantial improvements in sample efficiency and stronger overall performance in robotic manipulation tasks.
Vision-Language-Action (VLA) models have shown strong potential for general-purpose robotic manipulation, but their reliance on expert demonstrations limits their ability to learn from failures and perform self-corrections. Reinforcement learning (RL) addresses these through self-improving interactions with the physical environment, but suffers from high sample complexity on real robots. We introduce World-Model-based Policy Optimization (WMPO), a principled framework for on-policy VLA RL without interacting with the real environment. In contrast to widely used latent world models, WMPO focuses on pixel-based predictions that align the "imagined" trajectories with the VLA features pretrained with web-scale images. Crucially, WMPO enables the policy to perform on-policy GRPO that provides stronger performance than the often-used off-policy methods. Extensive experiments in both simulation and real-robot settings demonstrate that WMPO (i) substantially improves sample efficiency, (ii) achieves stronger overall performance, (iii) exhibits emergent behaviors such as self-correction, and (iv) demonstrates robust generalization and lifelong learning capabilities.