E0: Enhancing Generalization and Fine-Grained Control in VLA Models via Tweedie Discrete Diffusion

TL;DR

E0 introduces a discrete diffusion approach for VLA models that improves generalization, fine-grained control, and robustness across diverse robotic tasks and environments.

Contribution

The paper proposes E0, a novel Tweedie discrete diffusion framework that aligns with token-based reasoning and enhances action generation in VLA models.

Findings

Achieves state-of-the-art performance on 14 environments.

Outperforms baselines by 10.7% on average.

Enhances robustness to camera shifts with augmentation.

Abstract

Vision-Language-Action (VLA) models offer a unified framework for robotic manipulation by integrating visual perception, language understanding, and control generation. However, existing VLA systems still struggle to generalize across diverse tasks, scenes, and camera viewpoints, and often produce coarse or unstable actions. We argue that these limitations are closely tied to the structural properties of actions in VLA settings, including the inherent multi-peaked nature of action distributions, the token-based symbolic reasoning of pretrained VLM/VLA backbones, and the effective finite resolution imposed by real-world robotic control. Motivated by these properties, we introduce E0, a tweedie discrete diffusion framework that formulates action generation as iterative denoising over quantized action tokens. By operating in a discrete action space with a principled diffusion process, E0 naturally aligns with token-based reasoning, supports fine-grained yet executable action control, and avoids the distributional mismatch of masking-based discrete diffusion. We further introduce a spherical viewpoint perturbation augmentation to enhance robustness to camera shifts without additional data. Experiments on LIBERO, VLABench, ManiSkill, and a real-world Franka arm demonstrate that E0 achieves state-of-the-art performance across 14 diverse environments, outperforming strong baselines by 10.7% on average.