ACoT-VLA: Action Chain-of-Thought for Vision-Language-Action Models

TL;DR

This paper introduces ACoT-VLA, a novel vision-language-action model that uses structured action reasoning to improve robot manipulation, outperforming previous indirect reasoning methods.

Contribution

It proposes a new ACoT paradigm with explicit and implicit action reasoners, enabling more effective and grounded policy learning for robotic tasks.

Findings

ACoT-VLA outperforms existing models in real-world and simulation environments.

The explicit action reasoner generates coarse reference trajectories for better guidance.

The implicit reasoner extracts latent action priors to enhance policy grounding.

Abstract

Vision-Language-Action models have emerged as essential generalist robot policies for diverse manipulation tasks, conventionally relying on directly translating multimodal inputs into actions via Vision-Language Model embeddings. Recent advancements have introduced explicit intermediary reasoning-such as sub-task prediction (language) or goal image synthesis (vision)-to guide action generation. However, these intermediate reasoning are often indirect and inherently limited in their capacity to convey the full, granular information required for precise action execution. Instead, we posit that the most effective form of reasoning is one that deliberates directly in the action space. We introduce Action Chain-of-Thought (ACoT), a paradigm where the reasoning process itself is formulated as a structured sequence of coarse action intents that guide the final policy. In this paper, we propose ACoT-VLA, a novel architecture that materializes the ACoT paradigm. Specifically, we introduce two complementary components: an Explicit Action Reasoner (EAR) and Implicit Action Reasoner (IAR). The former proposes coarse reference trajectories as explicit action-level reasoning steps, while the latter extracts latent action priors from internal representations of multimodal input, co-forming an ACoT that conditions the downstream action head to enable grounded policy learning. Extensive experiments in real-world and simulation environments demonstrate the superiority of our proposed method. Code is available at: https://github.com/AgibotTech/ACoT-VLA.