发表机构
Nanyang Technological University; Satellite Research Centre (SaRC), Nanyang Technological University (NTU)（南洋理工大学; 南洋理工大学卫星研究中心）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
提出一种可达性感知制导架构，通过解析安全起始区域和模型预测控制，实现翻滚非合作目标自主接近，兼顾旋转视线约束与实时性。
AI 中文摘要
本文提出一种可达性感知制导架构，用于在旋转视线（LOS）对接走廊约束下自主接近翻滚的非合作目标。LOS允许集随目标本体坐标系旋转，在追踪器相对坐标中产生时变多面体约束。通过两个保守准则构建安全起始区域：(i) 方向性逐约束侵蚀，即推力能够抑制旋转引起的漂移之前所消耗的裕度；(ii) 同步范围界限 $r < 2a_{\max}/\omega_t^2$，确保追踪器能够抵消表观旋转速度而不超过保持点。闭环制导采用滚动时域MPC控制器，使用Clohessy-Wiltshire-Hill (CWH)预测动力学和二次规划中的显式LOS走廊约束。真实传播使用精确离散CWH状态转移矩阵并采用子步进，因此可行性声明是物理诚实的：不应用参考混合或状态投影。三区域跟踪律管理从远程惯性接近到本体坐标系共旋转和同步保持的过渡。将解析安全起始区域与四种标准可达性引擎（后向和前向多面体可达集、Hamilton-Jacobi水平集和闭环蒙特卡洛）进行基准测试：闭式准则比Hamilton-Jacobi可达性快250倍，同时在500个案例扫描中预测闭环可行性的精确度为0.80，召回率为0.91。剩余的6%假阳性率和与Hamilton-Jacobi的IoU差距量化了一个结构性质：同步集（到达和共旋转）是位置可达集的严格子集，该差距随翻滚率增大而增大。因此，解析边界是机载决策（go/no-go）的可靠内部证书，而Hamilton-Jacobi计算成本过高。
英文摘要
This paper presents a closed-form test that decides, before the maneuver begins, whether a chaser can reach and hold station at the hold point of a tumbling, uncooperative target inside a line-of-sight (LOS) corridor that turns with the target. A closed-loop controller cannot answer this question: a receding-horizon controller checks the corridor only over its prediction horizon, and near a tumbling target the corridor sweeps past faster than bounded thrust can follow, so a start that is feasible at the first step can become unrecoverable later. Today the only ways to know are to fly the controller in simulation or to compute a Hamilton--Jacobi reachable set, both too slow onboard. The test combines two criteria derived from bounded-thrust relative orbital dynamics: a directional erosion margin, the corridor margin that rotation-induced drift consumes before the thruster arrests it, and a synchronization radius, beyond which the apparent rotational velocity cannot be cancelled. Guidance pairs a three-regime tracking law with a receding-horizon quadratic program. Benchmarked against polytopic backward and forward reachable sets, Hamilton--Jacobi level sets and closed-loop Monte Carlo simulation, the test runs over two orders of magnitude faster than Hamilton--Jacobi and, over 500 closed-loop cases, predicts feasibility with 91% recall and 80% precision. The gap to Hamilton--Jacobi is structural, not a method error: reaching the hold point and co-rotating with it is a stronger requirement than arriving with arbitrary velocity, and the gap widens with tumble rate. The test therefore gives an onboard go/no-go answer where Hamilton--Jacobi reachability is too expensive.
Comments11 pages, 3 figures, 3 tables. Preprint of paper IAC-26,C1,3,6,x110087 accepted for publication at the 77th International Astronautical Congress (IAC 2026), Antalya, T"urkiye, 5--9 October 2026
Journal refProceedings of the 77th International Astronautical Congress (IAC 2026), Antalya, Türkiye, 2026
