ROAILGOct 18, 2025

Do What You Say: Steering Vision-Language-Action Models via Runtime Reasoning-Action Alignment Verification

NVIDIA
arXiv:2510.16281v111 citationsh-index: 34
Originality Incremental advance
AI Analysis

This work addresses a critical reliability issue in robotic instruction-following by improving robustness to out-of-distribution perturbations without retraining, which is incremental but impactful for deploying AI in real-world environments.

The paper tackles the problem of reasoning-action misalignment in vision-language-action models for robotics, where correct textual plans can still lead to incorrect actions, especially in out-of-distribution scenarios. It introduces a training-free runtime method that verifies alignment between plans and actions, achieving up to 15% performance gain on behavior composition tasks.

Reasoning Vision Language Action (VLA) models improve robotic instruction-following by generating step-by-step textual plans before low-level actions, an approach inspired by Chain-of-Thought (CoT) reasoning in language models. Yet even with a correct textual plan, the generated actions can still miss the intended outcomes in the plan, especially in out-of-distribution (OOD) scenarios. We formalize this phenomenon as a lack of embodied CoT faithfulness, and introduce a training-free, runtime policy steering method for reasoning-action alignment. Given a reasoning VLA's intermediate textual plan, our framework samples multiple candidate action sequences from the same model, predicts their outcomes via simulation, and uses a pre-trained Vision-Language Model (VLM) to select the sequence whose outcome best aligns with the VLA's own textual plan. Only executing action sequences that align with the textual reasoning turns our base VLA's natural action diversity from a source of error into a strength, boosting robustness to semantic and visual OOD perturbations and enabling novel behavior composition without costly re-training. We also contribute a reasoning-annotated extension of LIBERO-100, environment variations tailored for OOD evaluation, and demonstrate up to 15% performance gain over prior work on behavior composition tasks and scales with compute and data diversity. Project Website at: https://yilin-wu98.github.io/steering-reasoning-vla/

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes