外观
论文arXiv 2610.2047414 天前
AdaRSI：面向具身智能体的按需递归自我改进
原文链接：AdaRSI: On-Demand Recursive Self-Improvement for Embodied Agents（arXiv 2610.20474 · 发布日期：2026-10-14）
AdaRSI: On-Demand Recursive Self-Improvement for Embodied Agents
针对预训练VLA在未见长程任务中途失败，AdaRSI按需归因并微调不可靠子步骤，在RoboCasa365上验证提升成功率。
核心要点
- 预训练VLA常在未见长程任务中途失败，现有RSI难区分调用方式与执行能力问题。
- AdaRSI先改进技能与harness，再从保存状态重试失败子步骤以归因持续失败。
- 框架仅微调仍不可靠的子步骤，使用自收集示范并继承早期任务经验。
- RoboCasa365的50个目标任务上，未见组合任务成功率由4.6%升至13.8%。
验证范围与局限：验证范围为RoboCasa365全部50个目标任务，执行器为GR00T N1.5，且未使用人工示范。
英文摘要 (Abstract)
Pretrained vision-language-action models (VLAs) often fail midway through unseen long-horizon tasks. Recursive self-improvement (RSI) offers a remedy, but embodied RSI methods either refine the harness around a frozen executor and so cannot add a missing motion, or train the executor on persistent failures without asking whether a failure lies in how the executor is called or in what it can do. We ask when a self-improving embodied system should change the weights of its executor, and propose AdaRSI, an on-demand RSI framework for embodied agents. A coding agent first improves its skills and harness, then attributes persistent failures by retrying failing sub-steps from saved states, and fine-tunes only the sub-steps that remain unreliable, with demonstrations it collects itself and experience carried over from earlier tasks. On all 50 target tasks of RoboCasa365, with GR00T N1.5 as the executor and no human demonstrations, AdaRSI improves the VLA on all three suites and raises its success on unseen composite tasks from 4.6% to 13.8%. Compared with fine-tuning every sub-step, it reaches slightly higher success while using only 28% of the GPU time and 22% of the self-collected demonstrations, and fine-tuning without a harness trains more yet gains far less.
作者：Peidong Wang, Ying Chang, Zhiming Ma, Xiaocui Yang, Shi Feng, Daqi Liu, Yuzhe Yang, Xu Cao, Chuang Wang, Jian Su
来源
本页事实对应的公开页面，以原文为准。