# Evo-RL：从选择性经验中迭代机器人强化学习

> 原文链接：https://robotics.hellobike.com/papers/2026-10-09-2610-13345
> 频道：papers | 更新日期：2026-10-09 | 标签：arxiv, cs.RO
> 核心摘要：Evo-RL结合人工干预与优势引导的经验选择，在真实高精度操作任务上验证并提升机器人策略。

# Evo-RL：从选择性经验中迭代机器人强化学习

::: v-pre
> 原文链接：[Evo-RL: Iterative Robot Reinforcement Learning from Selective Experience](https://arxiv.org/abs/2610.13345)（arXiv 2610.13345 · 发布日期：2026-10-09）

**Evo-RL: Iterative Robot Reinforcement Learning from Selective Experience**

Evo-RL结合人工干预与优势引导的经验选择，在真实高精度操作任务上验证并提升机器人策略。

## 核心要点

- 现有真实世界强化学习会累积价值不一的经验，成功轨迹中也可能含低效或冗余行为。
- Evo-RL利用人工纠正和成功轨迹中估计优势为正的经验，迭代细化机器人策略。
- 作者引入RW-RL-Dataset，包含跨四种机器人形态的1000多小时经验以支持学习。
- 在真实世界高精度操作任务上，Evo-RL相较RECAP将尝试成功率提高30.46%，并公开代码、模型与数据集。

**验证范围与局限**：验证范围包括真实世界高精度操作任务，数据集覆盖四种机器人形态。

## 英文摘要 (Abstract)

Existing real-world reinforcement learning methods improve robot policies by combining human corrections with autonomous rollouts. However, accumulated experience varies in its value for policy improvement, and even successful trajectories can contain inefficient or redundant behavior, motivating more selective experience reuse. To this end, we propose Evo-RL, an iterative reinforcement learning method that combines human interventions with advantage-guided experience selection. Evo-RL uses human corrections and experience with positive estimated advantages from successful trajectories to iteratively refine the policy. To support learning from these different experience sources, we introduce RW-RL-Dataset, which spans over 1,000 hours across four robot embodiments. On a real-world high-precision manipulation task, Evo-RL improves attempt success rate over RECAP by 30.46\%. Our code, model checkpoints, and dataset are publicly available.

作者：Kai Ye, Yifan Jin, Qianhang Luo, Hongru Zhang, Jiedong Zhang, Shuyuan Hu, Song-Yuan Wu, Jihai Xu, Jie Zhao, Bo Zhao
:::

## 原始参考来源 (Sources)

- [arXiv 2610.13345：Evo-RL: Iterative Robot Reinforcement Learning from Selective Experience](https://arxiv.org/abs/2610.13345)