# 小错误攸关之处：量化模仿学习中的状态关键性

> 原文链接：https://robotics.hellobike.com/papers/2026-10-13-2610-17709
> 频道：papers | 更新日期：2026-10-13 | 标签：arxiv, cs.RO
> 核心摘要：针对模仿学习中状态重要性不均，SCRIM估计动作微扰的最坏价值退化，并在仿真与真实操作任务验证。

# 小错误攸关之处：量化模仿学习中的状态关键性

::: v-pre
> 原文链接：[Where Small Mistakes Matter: Quantifying State Criticality for Imitation Learning](https://arxiv.org/abs/2610.17709)（arXiv 2610.17709 · 发布日期：2026-10-13）

**Where Small Mistakes Matter: Quantifying State Criticality for Imitation Learning**

针对模仿学习中状态重要性不均，SCRIM估计动作微扰的最坏价值退化，并在仿真与真实操作任务验证。

## 核心要点

- 动机是行为克隆通常同等对待所有示范状态，但少数关键状态的小偏差可能导致失败。
- 方法将状态关键性形式化为有界专家动作扰动造成的最坏下游价值退化。
- SCRIM通过围绕固定模仿策略训练对抗扰动策略和值评论器，为示范状态打分。
- 实验显示其可重加权行为克隆、选择紧凑高价值数据子集，并提供关键时刻解释。

**验证范围与局限**：验证范围为多种仿真和真实世界机器人操作任务。

## 英文摘要 (Abstract)

Imitation learning typically treats every demonstration state as equally important, optimizing a behavior-cloning objective uniformly across the dataset. Yet not all states are created equal: in a small number of critical states, even a slight deviation from the expert action can lead to a failure, while in many other states such errors are recoverable and inconsequential. We formalize state criticality in imitation learning as the worst-case degradation in downstream value induced by a bounded perturbation of the expert action at a given state. We then propose SCRIM, a practical State CRIticality Measure that estimates this quantity by training an adversarial disturbance policy and a value critic around a fixed imitation policy. SCRIM assigns each demonstration state a score reflecting how vulnerable the task outcome is to small, policy-relevant action errors. We show that these scores can be used to reweight behavior-cloning losses, improving imitation at consequential states and increasing overall task performance. We further show that SCRIM enables data curation by selecting compact, high-value subsets of demonstrations that preserve performance while substantially reducing the amount of training data. We evaluate SCRIM across various simulated and real-world robotic manipulation tasks, and demonstrate that SCRIM improves policy training, enables efficient data selection, and provides interpretable insights into which moments in a demonstration are most critical for task success. Project website: https://sia-lab-git.github.io/SCRIM/

作者：Yusuf Umut Ciftci, Joonwon Kang, Ajay Suresha Sathya, Somil Bansal
:::

## 原始参考来源 (Sources)

- [arXiv 2610.17709：Where Small Mistakes Matter: Quantifying State Criticality for Imitation Learning](https://arxiv.org/abs/2610.17709)