# 超越动作：为 Vision-Language-Action 模型学习未来操作目标

> 原文链接：https://robotics.hellobike.com/papers/2026-10-09-2610-13212
> 频道：papers | 更新日期：2026-10-09 | 标签：arxiv, cs.RO
> 核心摘要：针对VLA仅预测动作难表征结果的问题，AFCE学习未来Effect，并在DexJoCo上验证提升成功率。

# 超越动作：为 Vision-Language-Action 模型学习未来操作目标

::: v-pre
> 原文链接：[Beyond Actions: Learning Future Operation Targets for Vision-Language-Action Models](https://arxiv.org/abs/2610.13212)（arXiv 2610.13212 · 发布日期：2026-10-09）

**Beyond Actions: Learning Future Operation Targets for Vision-Language-Action Models**

针对VLA仅预测动作难表征结果的问题，AFCE学习未来Effect，并在DexJoCo上验证提升成功率。

## 核心要点

- 动作只描述执行命令而不说明将实现的结果，VLA通常仍以动作块预测为主。
- AFCE以未来物理Effect为主要预测目标，由离线tokenizer构建有序Effect tokens。
- 流匹配VLA根据当前图像、指令和机器人状态预测Effect，再解码为动作块。
- 在DexJoCo上总体成功率50.97%，优于π0.5和UVT；项目页已公开。

**验证范围与局限**：验证范围为DexJoCo，摘要明确报告了五个双臂任务表现。

## 英文摘要 (Abstract)

Actions specify what a robot should execute, but do not describe what those commands will achieve. Vision-language-action (VLA) policies nevertheless learn primarily by predicting action chunks from images, language, and state. We introduce Action--Future Coupled Effect Learning (AFCE), which makes a future-grounded physical Effect the policy's primary prediction target. An offline tokenizer constructs temporally ordered Effect tokens from demonstrated actions, realized robot responses, and sparse visual transitions. Reconstruction objectives encourage the representation to retain executable detail and observed physical change. A flow-matching VLA predicts Effects from current images, instruction, and robot state, while a state-conditioned decoder converts them into action chunks. Embodiment-specific interfaces allow single- and dual-arm tasks to share the Effect model. On DexJoCo, AFCE achieves $50.97\%$ overall success, improving over $π_{0.5}$ and UVT by 11.82 and 15.21 percentage points, respectively. It leads on all five bimanual tasks, highlighting the value of Effect supervision for coordinated dexterous manipulation. The project page is available at https://afceproject.github.io/AFCE-Web/.

作者：Min Lin, Bokai Zhou, Jianqi Lin, Kaiyuan Yang, 昌懿 成, Junjie Yu, Xiaofan Liu, Jianhua Han, Rui Li, Shen Zhao
:::

## 原始参考来源 (Sources)

- [arXiv 2610.13212：Beyond Actions: Learning Future Operation Targets for Vision-Language-Action Models](https://arxiv.org/abs/2610.13212)