SCULPT-VLA: Learning Structured Control through Staged Action Grounding

2026-09-19

AI Summary

本文提出SCULPT-VLA方法，通过分阶段动作基础学习结构化控制，解决了视觉-语言-动作策略中如何有效利用中间表示的问题。

Abstract

Vision-language-action (VLA) policies increasingly incorporate structured intermediate supervision beyond action labels. Yet specifying what an intermediate representation should encode leaves open how action prediction learns to depend on it. We introduce \textbf{SCULPT-VLA}, a policy that learns structured control through staged action grounding. Its action-conditioning state comprises complementary factors for task progression, scene dynamics, and spatial grounding. Training first forms these factors with teacher scaffolds, then grounds coarse action prediction through their composition as scaffold inputs are withdrawn. Direct perceptual access is subsequently restored for continuous refinement, combining the learned state with perceptual detail. The curriculum separates learning to condition actions on structure from refining continuous control. Deployment requires neither teachers nor discrete-action autoregression. SCULPT-VLA achieves higher average success than shared-backbone baselines on LIBERO, SimplerEnv-WidowX, and RoboTwin 2.0 Full. On SimplerEnv-WidowX, final success is 83.5\%, versus 71.3\% when Stage-II action learning directly accesses vision and language. Across four physical robot tasks, average success under the tested distribution shifts reaches 58.1\%, compared with 45.6\% for $π_{0.5}$. Training ablations and factor-wise interventions support the staged design and show that the learned state continues to contribute to control after direct perceptual access is restored.

Problem

• structured control
• staged action grounding
• VLA policies
• intermediate representation
• action prediction

Innovation

• staged action grounding
• structured control
• curriculum learning
• state factorization