Efficient Reinforcement Learning for Satellite Evasion Tasks in Limited Observable and Orbit-Constrained Environments
Space: Science & Technology · 01.01.2026 · 6
DOI: 10.34133/space.0501 ↗Kurz gefasst
Fachartikel in der Zeitschrift Space: Science & Technology, online erschienen am 05.08.2026. Verfasst von Linsen Wei und 8 weiteren Autor:innen. Die Arbeit nennt 61 Literaturangaben. Gefördert von: National Natural Science Foundation of China, the Shanghai Natural Science Foundation, the Research Initiation Fund of NPU.
Abstract (Originalsprache)
Reinforcement learning (RL) holds promise for the automated control of satellite orbits and the avoidance of spatial obstacles due to its ability to learn and adapt to dynamic environments through interaction, enabling autonomous, real-time decision-making for orbit adjustments. Satellite orbit control tasks manifest strong nonlinear characteristics, for which algorithms grounded on neural networks are more adept at handling. However, RL methods based on neural networks, which are highly nonconvex and nonsmooth functions, sometimes yield extreme action values that are not conducive to the orbit-constrained task at hand. In addition, in practical satellite task environments, limited observability is often faced. This limited observability not only increases the complexity of the algorithm but also affects the efficiency of identifying appropriate actions. To address this issue, we introduce a hierarchical parameterization mechanism to accelerate the search for appropriate actions. Hence, we introduce the integration of self-attention mechanisms with the Kalman filter into RL frameworks to accommodate scenarios with limited observability. Experimental results demonstrate a marked increase in the success rate of automated evasion maneuvers in environments with limited observability and orbital constraints, when facing pursuing satellites. Then, we performed a comparative analysis, applying Lambert guidance and RL-based guidance as control strategies for powerful confinement environment to validate our algorithm’s robustness.
Förderung
- National Natural Science Foundation of China
- the Shanghai Natural Science Foundation
- the Research Initiation Fund of NPU
Zitieren
APA
Wei, L., Ning, X., Lian, X., Yang, K., Ni, T., Chen, F., Tang, R., Chen, Z., & Bo, L. (2026). Efficient Reinforcement Learning for Satellite Evasion Tasks in Limited Observable and Orbit-Constrained Environments. Space: Science & Technology, 6. https://doi.org/10.34133/space.0501
Vancouver
Wei L, Ning X, Lian X, Yang K, Ni T, Chen F, et al. Efficient Reinforcement Learning for Satellite Evasion Tasks in Limited Observable and Orbit-Constrained Environments. Space: Science & Technology. 2026;6. doi:10.34133/space.0501
BibTeX
@article{Wei2026,
author = {Wei, Linsen and Ning, Xin and Lian, Xiaobin and Yang, Kaiyuan and Ni, Tao and Chen, Feng and Tang, Rugang and Chen, Zhansheng and Bo, Liteng},
title = {Efficient Reinforcement Learning for Satellite Evasion Tasks in Limited Observable and Orbit-Constrained Environments},
journal = {Space: Science & Technology},
year = {2026},
volume = {6},
doi = {10.34133/space.0501}
}
EUVIMED