Article

Connecting VLA and world-action models with a unified model-to-policy formal framework

The paper proposes an architecture-agnostic probabilistic formal framework based on stochastic kernels for uniformly describing embodied foundation models. The framework has two parts: characterizing models with conditional output distributions and action-realization kernels to induce task-conditioned robot policies, and connecting pretraining, post-training, and policy deployment. The authors instantiate it on OpenVLA and DreamZero, covering both autoregressive action-token prediction and joint video-action generation paradigms.

Part of the event: Proposes a Unified Formal Framework for Embodied Foundation Models

This page provides a summary and source link only.