# Reflex：视觉扎根的反应式人形机器人控制

> 原文链接：https://robotics.hellobike.com/papers/2026-10-08-2610-12772
> 频道：papers | 更新日期：2026-10-08 | 标签：arxiv, cs.RO
> 核心摘要：Reflex 面向人形机器人接抛掷箱子，分三阶段学习视觉反应式全身控制，并在仿真和真实 Unitree G1 上验证。

# Reflex：视觉扎根的反应式人形机器人控制

::: v-pre
> 原文链接：[Reflex: Visually Grounded Reactive Humanoid Control](https://arxiv.org/abs/2610.12772)（arXiv 2610.12772 · 发布日期：2026-10-08）

**Reflex: Visually Grounded Reactive Humanoid Control**

Reflex 面向人形机器人接抛掷箱子，分三阶段学习视觉反应式全身控制，并在仿真和真实 Unitree G1 上验证。

## 核心要点

- 视觉反应式全身控制需协调全身、实时应对物体动态并接地自我中心视觉。
- Reflex 先以强化学习训练基于状态的全身接箱策略。
- 随后从延迟不完整观测推断箱子动态，并由自我中心 RGB-D 恢复该表示。
- 仿真中 RGB-D 成功率为 85.7%；真实 Unitree G1 在 2.5 和 5 m 成功率为 65% 和 40%。

**验证范围与局限**：验证范围为仿真接箱任务和一台真实 Unitree G1 接人类抛掷箱子。

## 英文摘要 (Abstract)

Humans effortlessly react to the world; they adeptly return a tennis shot, catch a falling cup, step aside from a passing cart, or pass a box tossed during truck unloading. These everyday actions require anticipating a moving target and reflexively responding with the whole body within about a second. Such visual reactive whole-body control remains challenging for humanoids, which must (1) coordinate whole-body control, (2) react to object dynamics in real time, and (3) ground perception in egocentric vision. We study catching thrown boxes as a representative task that demands all three capabilities; existing systems cover only part of it, relying on end-effectors, motion capture, or marked objects. We present Reflex, a three-stage framework that learns these capabilities stage by stage while connecting them through a compact representation of object dynamics. First, a state-based policy learns whole-body catching with reinforcement learning. Second, the policy learns to infer the dynamics of a moving box from delayed and incomplete observations. Third, an egocentric RGB-D encoder learns to recover this dynamics representation directly from visual observations while keeping the whole-body controller fixed. This decomposition provides each capability with a direct training signal and makes visual learning substantially more scalable. In simulation, Reflex achieves 85.7% catch success from egocentric RGB-D, compared with 87.9% when given privileged box observations. Temporal visual history is critical: removing it reduces success by 35.6 and 39.8 points on centered and laterally offset throws. On a real Unitree G1, Reflex catches human-thrown boxes with 65% and 40% success at 2.5 and 5 m, reaching 76% and 59% of human catchers' success at the same distances.

作者：Taoyang Jia, Weikai Huang, Linxin Song, Jared Darlington, Jieyu Zhao, Yue Wang, Jiafei Duan, Zhongzheng Ren, Ranjay Krishna
:::

## 原始参考来源 (Sources)

- [arXiv 2610.12772：Reflex: Visually Grounded Reactive Humanoid Control](https://arxiv.org/abs/2610.12772)