Scholarlink

This paper proposes a reinforcement learning-based control framework for ensuring the safety of sensitive payloads during spacecraft attitude adjustments. The framework specifically addresses the challenge of preventing payload exposure to obstacle areas, such as intense radiation fields or high-intensity light sources. By implementing the Soft Actor-Critic (SAC) algorithm with pointing constraints embedded in the reward function, the system achieves autonomous hazard avoidance while executing attitude maneuvers. Furthermore, compared to the conventional method’s, via optimizing energy consumption coefficients, the framework reduces propellant consumption from 2940 mg to 340 mg, This reinforcement learning framework provides an efficient and reliable solution for attitude control of jet-driven spacecraft during attitude adjustments with pointing constraints.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Reinforcement Learning-Based Spacecraft Attitude Control Under Pointing Constraints

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Yuxiang Luo,
  • Yujia E,
  • Desong Du,
  • Yuquan Li,
  • Yanfang Liu

摘要

This paper proposes a reinforcement learning-based control framework for ensuring the safety of sensitive payloads during spacecraft attitude adjustments. The framework specifically addresses the challenge of preventing payload exposure to obstacle areas, such as intense radiation fields or high-intensity light sources. By implementing the Soft Actor-Critic (SAC) algorithm with pointing constraints embedded in the reward function, the system achieves autonomous hazard avoidance while executing attitude maneuvers. Furthermore, compared to the conventional method’s, via optimizing energy consumption coefficients, the framework reduces propellant consumption from 2940 mg to 340 mg, This reinforcement learning framework provides an efficient and reliable solution for attitude control of jet-driven spacecraft during attitude adjustments with pointing constraints.