具身开发者社区

外观

论文arXiv 2609.264671 个月前

通才行动，专家介入：用于视觉-语言-动作操作的模块化阶段选择性强化学习 ​

针对多阶段操作精密阶段VLA表现不足，RouteRLT用阶段感知协调对接VLA与RL专家，并在LIBERO系列和真实连接器任务验证。

AI 整理
  • 本体 SDK 与控制
  • 具身模型
  • 评测与安全

原文链接：Generalists Act, Specialists Intervene: Modular Stage-Selective Reinforcement Learning for Vision-Language-Action Manipulation（arXiv 2609.26467 · 发布日期：2026-09-22）

Generalists Act, Specialists Intervene: Modular Stage-Selective Reinforcement Learning for Vision-Language-Action Manipulation

针对多阶段操作精密阶段VLA表现不足，RouteRLT用阶段感知协调对接VLA与RL专家，并在LIBERO系列和真实连接器任务验证。

核心要点 ​

  • 多阶段操作中的精密关键阶段常使VLA模型受限，何时由RL专家接管仍是开放问题。
  • RouteRLT以通用VLA为默认控制器，配合指定精密阶段RL专家，学习阶段感知交接机制。
  • 在LIBERO上提高成功率，在LIBERO-Plus上保持净收益，显示协调能增强任务完成。
  • 真实连接器拾取与插入任务中，RouteRLT完成65.7%试次，基线为8.6%。

验证范围与局限：验证范围包括LIBERO、LIBERO-Plus基准和一个真实连接器拾取与插入任务。

英文摘要 (Abstract) ​

Vision-language-action (VLA) models often struggle in the precision-critical phases of multi-stage manipulation tasks. To mitigate this issue, VLA models can be used in conjunction with reinforcement learning (RL) specialists that are specifically trained to handle the precision-critical phases. However, the coordination between the base VLA model and the RL specialists, which dictates when a specialist should take over from the base VLA and vice-versa, remains an open research question. In this paper, we address this gap by introducing RouteRLT, a modular framework that coordinates a generalist VLA, used as the default controller, with designated precision-critical RL specialists. At a high level, our framework trains a phase-aware coordination mechanism that handles handoffs between the generalist and the specialists. We evaluate RouteRLT on the LIBERO and LIBERO-Plus benchmarks, as well as on a physical connector pickup and insertion task. Overall, we find that RouteRLT improves success on LIBERO, and retains net gains on LIBERO-Plus. On the physical task, RouteRLT completes 65.7% of trials, compared with 8.6% for the baseline. Altogether, these results demonstrate that learned coordination builds on generalist VLA capabilities to improve task completion in precision-critical manipulation.

作者：Chongyu Zhu, Hyegang Kim, Jaden Hinds, Juan Sebastián Rojas, Ramy ElMallah, Chi-Guhn Lee

来源

本页事实对应的公开页面，以原文为准。

  1. arXiv 2609.26467：Generalists Act, Specialists Intervene: Modular Stage-Selective Reinforcement Learning for Vision-Language-Action Manipulation访问 2026-10-21