# FRAMES：面向人形机器人移动操作的具身技能故障恢复与监控

> 原文链接：https://robotics.hellobike.com/papers/2026-09-18-2609-22538
> 频道：papers | 更新日期：2026-09-18 | 标签：arxiv, cs.RO
> 核心摘要：FRAMES面向人形移动操作执行失败，用VLM监控与恢复监督技能，并在MuJoCo五任务试验验证监控模块。

# FRAMES：面向人形机器人移动操作的具身技能故障恢复与监控

::: v-pre
> 原文链接：[FRAMES: Failure Recovery And Monitoring of Embodied Skills for Humanoid Loco-Manipulation](https://arxiv.org/abs/2609.22538)（arXiv 2609.22538 · 发布日期：2026-09-18）

**FRAMES: Failure Recovery And Monitoring of Embodied Skills for Humanoid Loco-Manipulation**

FRAMES面向人形移动操作执行失败，用VLM监控与恢复监督技能，并在MuJoCo五任务试验验证监控模块。

## 核心要点

- LLM规划器能分解指令并选择技能，但物理执行失败会破坏长程计划。
- 框架位于CEER全身控制器之上，由规划、VLM监控、恢复和记忆模块协同。
- 监控代理结合时间多视角观测、机器人与接触证据，并用深度和分割做几何 grounding。
- 在MuJoCo五项任务100次试验中，监控检出48/50失败、接受46/50成功，准确率94.0%。

**验证范围与局限**：验证仅独立评估了MuJoCo中的监控模块，完整恢复闭环端到端评估仍在进行。

## 英文摘要 (Abstract)

Large language model (LLM) planners can decompose natural-language instructions and select reusable robot skills, but choosing the correct skill does not guarantee successful physical execution. This gap is especially important in humanoid loco-manipulation, where errors during approach, grasping, transport, or placement can invalidate the remainder of a long-horizon plan. We present FRAMES, a failure-aware supervisory framework for the Unitree G1 humanoid that operates above the CEER whole-body controller. A Planner Agent selects subtasks through parameterized mid-level skills, while a vision-language-model-based Monitor Agent evaluates each skill using temporal multi-view observations and structured robot and contact evidence. Detected failures stop the active skill and provide grounded feedback to a Recovery Agent. The framework further includes a Memory Module for reusing prior skill experience, and geometric grounding via depth and segmentation. We independently evaluate the monitoring module of the framework in MuJoCo using 100 trials comprising 50 failed and 50 successful executions across five tasks. The monitor detects 48 of 50 failures, correctly accepts 46 of 50 successful executions, and achieves 94.0% overall accuracy. These results provide initial evidence for the monitoring component, while end-to-end evaluation of the complete recovery loop remains ongoing.

作者：Ajay Vikram Periasami, Xinyuan Luo, Haoyu Li, Xianyi Cheng
:::

## 原始参考来源 (Sources)

- [arXiv 2609.22538：FRAMES: Failure Recovery And Monitoring of Embodied Skills for Humanoid Loco-Manipulation](https://arxiv.org/abs/2609.22538)