Abstract
Many Vision-Language-Action (VLA) models rely on an internal world model trained via next-frame prediction. This approach, however, struggles with physical reasoning as it entangles static appearance with dynamic motion, often resulting in implausible visual forecasts and inefficient policy learning. To address these limitations, we introduce the Visual Chain of Thought (Visual CoT): a pre-training framework that encourages a model to reason about how a scene evolves before predicting what it will look like. We instantiate this principle in FlowVLA, which predicts a future frame ($v_{t+1}$) only after generating an intermediate optical flow representation ($f_t$) that encodes motion dynamics. This ``$v_t \rightarrow f_t \rightarrow v_{t+1}$'' reasoning process is implemented within a single autoregressive Transformer, guiding the model to learn disentangled dynamics. As a result, FlowVLA produces coherent visual predictions and facilitates more efficient policy learning. Experiments on challenging robotics manipulation benchmarks demonstrate state-of-the-art performance with substantially improved sample efficiency, pointing toward a more principled foundation for world modeling. Project page: this https URL
Abstract (translated)
许多视觉-语言-行动（VLA）模型依赖于通过下一帧预测训练的内部世界模型。然而，这种做法在物理推理方面存在困难，因为它将静态外观与动态运动纠缠在一起，通常会导致不可信的视觉预测和低效的行为学习。为了克服这些限制，我们引入了视觉思维链（Visual CoT）：一个预训练框架，鼓励模型先推断场景如何演变，然后再预测其未来的样子。我们将这一原则在FlowVLA中实现，在生成编码运动动态的中间光流表示($f_t$)之后才预测未来的帧($v_{t+1}$)。这种"$v_t \rightarrow f_t \rightarrow v_{t+1}$"推理过程在一个单一的自回归Transformer内部实施，指导模型学习解耦的动力学。因此，FlowVLA能够生成连贯的视觉预测，并促进更高效的行为学习。 在具有挑战性的机器人操作基准测试中的实验表明，与现有方法相比，FlowVLA表现出卓越的性能和显著提高的样本效率，这指向了一个更为原则化的世界建模基础。项目页面：[此网址](https://this-url.com/)
URL
https://arxiv.org/abs/2508.18269
