国际空间站上首次在轨演示深度学习热启动加速轨迹优化
Deep Learning Warm Starts for Trajectory Optimization on the International Space Station
AI 导读
研究人员在国际空间站的 Astrobee 自由飞行机器人上完成首次在轨演示，用神经网络为 SCP 求解器的轨迹优化提供热启动。该网络预测轨迹生成问题的解，由 SCP 求解器保证安全约束，使含旋转动力学情形的求解迭代次数减少 60%，含障碍物情形减少 50%，为航天领域的机器学习控制应用迈出重要一步。
正文 · 原文
Abstract:Trajectory optimization is a cornerstone of modern robot autonomy, enabling systems to compute trajectories and controls in real-time while respecting safety and physical constraints. However, it has seen limited usage in spaceflight applications due to its heavy computational demands that exceed the capability of most flight computers. In this work, we provide results on the first in-space demonstration of using machine learning-based warm starts for accelerating trajectory optimization for the Astrobee free-flying robot onboard the International Space Station (ISS). We formulate a data-driven optimal control approach that trains a neural network to learn the structure of the trajectory generation problem being solved using sequential convex programming (SCP). Onboard, this trained neural network predicts solutions for the trajectory generation problem and relies on using the SCP solver to enforce safety constraints for the system. Our trained network reduces the number of solver iterations required for convergence in cases including rotational dynamics by 60% and in cases with obstacles drawn from the training distribution of the warm start model by 50%. This work represents a significant milestone in the use of learning-based control for spaceflight applications and a stepping stone for future advances in the use of machine learning for autonomous guidance, navigation, & control.
| Comments: | International Conference on Space Robotics (iSpaRo) 2025. this https URL Earlier version presented at Workshop on Space Robotics at Robotics: Science and Systems (RSS) 2025 |
| Subjects: | Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) |
| Cite as: | arXiv:2505.05588 [cs.RO] |
| (or arXiv:2505.05588v4 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2505.05588 arXiv-issued DOI via DataCite |
|
| Journal reference: | 2025 International Conference on Space Robotics (iSpaRo), Sendai, Japan, 2025, pp. 394-401 |
| Related DOI: | https://doi.org/10.1109/iSpaRo66239.2025.11436159
DOI(s) linking to related resources |
Submission history
From: Somrita Banerjee [view email]
[v1]
Thu, 8 May 2025 18:42:36 UTC (328 KB)
[v2]
Wed, 13 Aug 2025 01:24:55 UTC (3,894 KB)
[v3]
Tue, 4 Nov 2025 21:15:31 UTC (4,349 KB)
[v4]
Thu, 22 Oct 2026 18:24:37 UTC (4,349 KB)
来源：arXiv cs.AI · arxiv.org