外观
论文arXiv 2610.1922413 天前
月球车自主对接VLA策略：自适应执行地平线平衡算力与安全
本文由 AI 根据公开内容整理，可能有误，请以原文为准：End-to-End Vision-Language-Action Policies for Autonomous Lunar Rover Docking（arXiv 2610.19224）
核心内容
针对月表极端光照与感知扰动挑战，研究团队设计了用于自主对接的视觉-语言-动作（VLA）策略，以替代误差累积的传统模块化导航管线。该研究构建了可扩展训练框架，在高保真月面仿真环境中生成安全感知专家演示数据。闭环评估表明，该VLA策略在训练集未覆盖的感知扰动下，性能仍能与模块化基线持平，验证了互联网规模预训练模型在太空机器人领域的泛化潜力。
为解决星载算力受限与闭环安全的矛盾，论文提出自适应执行地平线机制。该策略根据环境开放度动态规划单次开环行驶距离，而非采用固定时间步长。实验证实，相比固定地平线方案，自适应机制在显著降低模型推理频率的同时，有效限制了危险接触率，实现了计算资源消耗与运行安全性的最优权衡，为地外探测器端侧部署提供了工程范式。
关键看点
- VLA月面适配：利用视觉-语言预训练常识推理能力，在有限任务数据下实现鲁棒泛化，闭环性能对标传统模块化管线
- 仿真数据引擎：构建可扩展训练框架，在高保真月面仿真中自动生成带安全约束的专家演示，解决地外真实样本匮乏难题
- 自适应执行地平线：动态调整开环控制时长以匹配环境风险，较固定步长显著降低推理频次并减少危险接触率
- 抗扰动验证：在训练分布外的感知干扰条件下完成闭环测试，验证端到端策略对月表极端光照与噪声的鲁棒性
工程与产业观察
该工作证明了VLA模型在非结构化地外环境的工程可行性，其自适应执行地平线机制为具身智能在算力受限边缘端的部署提供了关键参考。通过仿真生成安全感知数据弥补真实样本缺失，以及动态调节推理频率平衡安全与能耗的思路，对地球端野外机器人及特种作业机器人的算法架构设计具有直接借鉴价值。
英文摘要 (Abstract)
Surface navigation on the Moon is subject to extreme illumination contrast and deployment-time perceptual disturbances. While modular navigation pipelines have long supported planetary rover navigation, robustness is compromised by measurement errors that propagate through hand-engineered components. End-to-end learned policies are an alternative approach to navigation, which maps observations directly to control actions without intermediate components. Vision-language-action models, in particular, have driven significant breakthroughs in robotics and autonomous vehicle research. By leveraging internet-scale vision-language pretraining, these models can be exploited for their commonsense reasoning and visual understanding to provide robust generalization with limited task-specific data. However, their applicability and performance have yet to be characterized for lunar surface operations. In this study, we design a vision-language-action policy for autonomous lunar rover docking and characterize its performance relative to a modular navigation pipeline. We develop a scalable training framework that generates safety-aware expert demonstrations in a high-fidelity lunar simulation environment. In closed-loop evaluations, our policy performs competitively relative to the modular baseline under perceptual disturbances unseen during training. To address the tradeoff between compute consumption and closed-loop safety, the policy adapts its execution horizon by determining how far the rover may safely travel in open loop. Adaptive-horizon execution reduces inference frequency while limiting hazard-contact rates relative to fixed horizons in the evaluated scenarios.
作者：Sitan Huang, Grace Gao
来源
本页事实对应的公开页面，以原文为准。