arXiv cs.AI· Chlo\'e Carer (LGI), Adam Abdin (LGI)·· 1 天前AI 评分18
基于安全强化学习的在轨服务机会约束检查视点排序
Chance-Constrained Inspection Viewpoint Sequencing for On-Orbit Servicing via Safe Reinforcement Learning
AI 导读
该研究将航天器近距离检查的视点选择建模为约束马尔可夫决策过程（CMDP），用带拉格朗日松弛的 PPO 求解，并结合 CW 方程评估视点转移。在 Galileo 级导航卫星点云模型上，该方法在各风险预算下将禁飞区违规率控制在限定范围内，覆盖率接近最大值且不超 Δv 预算；相比固定禁飞半径加确定性规划，同等覆盖率下违规率不到一半。碰撞风险由此成为显式设计参数，无需逐案重新调参。
正文 · 原文
Abstract:Autonomous inspection of a target spacecraft is typically the initial step in on-orbit servicing. Planning close-proximity observation viewpoints involves a complex combinatorial problem constrained by coverage, fuel, and safety considerations. This paper models the viewpoint selection as a Constrained Markov Decision Process (CMDP) and applies Proximal Policy Optimization with Lagrangian relaxation to solve it. The system combines a high-level planner, which sequences camera viewpoints on an observation sphere around the target, with a low-level model that assesses each transfer using the Clohessy--Wiltshire (CW) equations for relative motion. The deterministic constraints on single-impulse reachability and $\Delta v$ cap are enforced directly on the action set. Safety margins are affected by navigation and manoeuvre errors, making the actual clearance stochastic. To address this, a chance constraint on the realised trajectory is implemented, with a Lagrangian multiplier governed by PID control to adapt the penalty during training. The approach is tested on a Galileo-class navigation satellite in medium Earth orbit, modelled as a point cloud where visibility from each viewpoint considers field-of-view and line-of-sight occlusion. Across various risk budgets, the trained policy maintains the keep-out violation rate within the specified limit, achieving coverage close to its maximum and staying within the mission $\Delta v$ budget. Compared to the usual practice of applying a fixed margin to the keep-out radius and planning deterministically, the chance-constrained policy attains less than half the violation rate at comparable coverage. This makes the collision risk an explicit design parameter, eliminating the need for re-tuning per case. The learned policy selects a viewpoint without the repeated Monte Carlo risk estimates required by a greedy, chance-constrained planner at every decision point.
| Subjects: | Artificial Intelligence (cs.AI); Optimization and Control (math.OC) |
| Cite as: | arXiv:2610.41639 [cs.AI] |
| (or arXiv:2610.41639v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.41639 arXiv-issued DOI via DataCite (pending registration) |
|
| Journal reference: | 77th International Astronautical Congress (IAC), International Astronautical Federation (IAF), Oct 2026, Antalya (TR), Turkey |
Submission history
From: Adam Abdin [view email] [via CCSD proxy]
[v1]
Fri, 30 Oct 2026 12:15:31 UTC (316 KB)
来源：arXiv cs.AI · arxiv.org