---
title: "When, Where, and What to Correct: Human-Guided Adaptive Residual Correction for Robot Policies"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2610.37547v1"
permalink: "https://feed.astralor.com/items/itm_5ea610c5e271d3617a6b9f12"
published: "2026-10-27T16:58:54.000Z"
score: 54
channel: "finance"
tags: ["RAG","When","Where","What","Correct","Human-Guided Adaptive Residual","Correction","Robot Policies Imitation"]
---

# When, Where, and What to Correct: Human-Guided Adaptive Residual Correction for Robot Policies

## At a glance

Imitation policies that reliably execute most of a manipulation trajectory can still fail in localized regions requiring last-mile precision. Updating the full policy or exploring residual actions everywhere treats this…

## Key points

- Imitation policies that reliably execute most of a manipulation trajectory can still fail in localized regions requiring last-mile precision.
- Updating the full policy or exploring residual actions everywhere treats this local failure as a global learning problem.
- We introduce Adaptive Residual Correction (ARC), a lightweight post-training module that uses reinforcement learning (RL) to learn a residual correction on top of a frozen policy while using sparse human corrections to …

## Full text

Imitation policies that reliably execute most of a manipulation trajectory can still fail in localized regions requiring last-mile precision. Updating the full policy or exploring residual actions everywhere treats this local failure as a global learning problem. We introduce Adaptive Residual Correction (ARC), a lightweight post-training module that uses reinforcement learning (RL) to learn a residual correction on top of a frozen policy while using sparse human corrections to structure autonomous exploration. Rather than treating human corrections as imitation targets, ARC uses pre-intervention states to determine 'when' residual control should become active and correction geometry to determine 'where' in residual action space autonomous refinement should explore; task reward determines 'what' residual is ultimately executed. Across three precision-manipulation tasks on a real robot, ARC improves average task success from 33.3% to 75.0%. The human corrections define task-dependent 2-3 dimensional subspaces of the seven-joint residual space that retain 96.0-97.7% of squared correction magnitude. Together, these results show that sparse human corrections can localize and constrain autonomous post-training without requiring the observed correction itself to serve as the final action target.

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_5ea610c5e271d3617a6b9f12