Agentic AutoResearch forSpace Autonomy: An Auditable, LLM-Driven Research Agent for Aerospace Control Problems
作者: Amit Jain, Richard Linares
分类: cs.RO, math.OC
发布日期: 2026-06-18
💡 一句话要点
提出AutoResearch框架以解决航天控制问题
🎯 匹配领域: 支柱九：具身大模型 (Embodied Foundation Models)
关键词: 航天控制 自主研究 大型语言模型 策略验证 噪声干扰
📋 核心要点
- 现有方法在航天器控制策略开发中面临噪声干扰和验证困难，导致结果的可靠性不足。
- 论文提出AutoResearch框架，利用大型语言模型自动化研究过程，并引入可信度层以验证结果的有效性。
- 在两个航天控制问题上，经过审计的策略显著超越了测量的种子噪声，展示了优越的性能和可靠性。
📝 摘要（中文）
随着航天器的引导、导航和控制功能越来越多地依赖于从专家求解器中提炼的学习策略，开发这样的策略本身就是一个研究过程。本文提出了AutoResearch框架，利用大型语言模型自主驱动航天控制问题的研究循环，并在循环中内置可信度层，验证每个报告结果是否符合问题的测量种子噪声。该语言模型仅作为离线研究代理，开发控制策略，生成的策略随后部署在航天器上，而模型本身并不直接操作航天器。每次迭代中，代理读取问题描述和运行历史，提出对训练脚本的单一编辑，执行并记录结果。只有在通过三项检查后，结果才会被认可。该方法在两个航天控制问题上应用，均显示出显著的性能提升。
🔬 方法详解
问题定义：本文旨在解决航天器控制策略开发中的噪声干扰和结果验证问题。现有方法往往难以区分真实改进与随机噪声，导致结果不可靠。
核心思路：AutoResearch框架通过大型语言模型自动化研究循环，结合可信度层来验证每个实验结果的有效性，确保所报告的改进是基于真实数据而非噪声。
技术框架：该框架包括多个模块：首先，代理读取问题描述和历史记录；其次，提出对训练脚本的单一编辑并执行；最后，记录结果并进行三项验证检查。
关键创新：最重要的创新在于引入了可信度层，确保每个结果在报告前经过严格验证，显著提高了结果的可信性和有效性。
关键设计：在参数设置上，模型通过测量种子噪声、重新验证最佳配置和逐一剔除代理编辑来进行结果验证，确保每一步的可靠性。
🖼️ 关键图片
📊 实验亮点
实验结果表明，经过审计的策略在两个航天控制问题上均显著超越了测量的种子噪声，特别是在对接问题上，未指导搜索未能生成可行策略，而学习的策略在每个种子上均保持在安全区域外，显示出明显的性能优势。
🎯 应用场景
该研究的潜在应用领域包括航天器的自主导航与控制、无人机飞行控制以及其他需要高可靠性的自动化系统。通过提高控制策略的可靠性，未来可在更复杂的环境中实现更安全的航天任务。
📄 摘要（原文）
Spacecraft guidance, navigation, and control functions are increasingly realized as learned policies distilled from expert solvers. Developing such a policy is itself a research process: an investigator selects an architecture and hyperparameters, runs experiments, and must determine whether an apparent improvement is genuine or merely seed noise. This paper presents AutoResearch, a framework in which a large language model autonomously drives that loop for aerospace control problems, coupled with a credibility layer, built into the loop, that certifies each reported result against the problem's own measured seed noise. The language model serves only as the offline research agent that develops the control policy; the trained policy it produces is then deployed onboard the spacecraft, while the model itself never operates the vehicle. At each iteration the agent reads a plain-language problem description and the run history, proposes a single edit to the training script, executes it, and logs the outcome. No reported result is credited until it passes the same three checks: measured per-problem seed noise, reseeded verification of the best configuration, and leave-one-out pruning of the agent's edits. The same loop is applied, unchanged, to two aerospace control problems: a Clohessy-Wiltshire relative rendezvous and a safety-constrained collision-avoidance docking past a keep-out zone, each calibrated against a known optimal control benchmark. In both, the audited policy clears the measured seed noise by many standard deviations; an undirected search over the same parameters does not. On the docking problem the gap becomes categorical: undirected search yields no feasible policy, while the learned policy stays outside the keep-out zone on every seed.