arXiv 2602.094302026-08-17cs.ROcs.AI

Sci-VLA：用于科学实验中长周期任务的代理VLA推理插件

AtomBridge: Agentic VLA Inference Plugin for Long-Horizon Tasks in Scientific Experiments

School of Computer Science and Engineering & School of Software Engineering & School of Artificial Intelligence, Southeast University, Nanjing, China（计算机科学与工程学院、软件工程学院、人工智能学院，东南大学，南京，中国）；Pattern Learning and Mining Lab, Southeast University, Nanjing, China（模式学习与挖掘实验室，东南大学，南京，中国）

Yiwen Pang, Bo Zhou, Changjin Li, Xuanhao Wang, Shengxiang Xu, Deng-Bao Wang, Peng Cheng, Shimin Di, Jingkuan Song, Min-Ling Zhang

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

本文提出了一种基于大语言模型的代理VLA推理插件，用于提升科学实验中长周期任务的执行效率和成功率。

摘要：

机器人实验室在自主科学发现中发挥着关键作用，通过实现可扩展的连续实验执行。最近的视觉-语言-动作（VLA）模型为机器人实验室提供了一个有前景的基础。然而，科学实验通常涉及由多个原子任务组成的长周期任务，这对现有VLA模型提出了根本性的挑战。虽然针对科学任务微调的VLA模型可以可靠地执行训练期间看到的原子实验动作，但它们往往无法执行由重新排列和组合这些已知原子动作形成的复合任务。这种限制源于训练时间的原子任务与推理时间的复合任务之间的分布不匹配，这阻止了VLA模型在原子任务之间执行必要的过渡操作。为了解决这一挑战，我们提出了一个用于科学实验中长周期任务的代理VLA推理插件。它引入了一个基于大语言模型（LLM）的代理推理机制，在执行顺序操纵任务时进行干预。通过执行显式的过渡推理并生成过渡性的机器人动作代码，所提出的插件引导VLA模型通过缺失的过渡步骤，使复合科学工作流程的可靠执行成为可能，而无需任何额外的训练。这种仅推理的干预使我们的方法在计算上高效、数据上高效，并且适合开放性及长周期的机器人实验室任务。我们构建了现有仿真环境中的3D科学仪器和常见科学操作场景。在这些场景中，我们验证了我们的方法在推理期间将每个原子任务的平均成功率提高了42%。此外，我们还表明，我们的方法可以轻松地从仿真转移到现实的科学实验室。

英文摘要：

Robotic laboratories play a critical role in autonomous scientific discovery by enabling scalable, continuous experimental execution. Recent vision-language-action (VLA) models offer a promising foundation for robotic laboratories. However, scientific experiments typically involve long-horizon tasks composed of multiple atomic tasks. Existing VLA models may fail to perform composed tasks formed by reordering and composing these known atomic actions. This limitation can arise from a skill-chaining gap caused by robot-state mismatch: the terminal robot state of one skill can fall outside the valid initial-state distribution of the next. To address this challenge, we propose AtomBridge, an Agentic VLA Inference Plugin for Long-Horizon Tasks in Scientific Experiments. AtomBridge attaches at inference time to a VLA policy already fine-tuned on atomic tasks, while keeping its weights fixed. At each task boundary, it uses LLM-based transition reasoning and robotic-action code generation to insert transitional actions between consecutive tasks. This plug-and-play design mitigates the skill-chaining gap caused by robot-state mismatch without additional VLA fine-tuning or demonstrations of composed long-horizon sequences. Across scientific manipulation sequences in simulation and a real-world experimental environment, AtomBridge improves execution continuity and per-step atomic-task success. On 8-step composed tasks, AtomBridge improves full-sequence success by 10%~25%.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。