课程驱动深度强化学习实现火箭远距离自主着陆
原标题：Curriculum-Driven Deep Reinforcement Learning for Autonomous Far-Range Rocket Landing
AI 摘要
该论文提出一种课程驱动的深度强化学习框架，用于可重复使用火箭的远距离导航与精确着陆。方法采用三阶段课程、critic 预热和排练机制防止灾难性遗忘，并使用基于完整旋转矩阵的高维观测与重新设计的奖励函数。在评估中，所提策略成功率达 99.98%，基线为 68.31%，且保留 69.1% 初始燃料，最终距着陆垫中心平均 7.07 单位，优于基线的 13.10 单位。
正文节选
Curriculum-Driven Deep Reinforcement Learning for Autonomous Far-Range Rocket Landing Abstract Autonomous vertical landing of reusable launch vehicles is a critical yet formidable challenge in aerospace engineering, primarily due to highly nonlinear dynamics, stringent precision requirements, and inherent environmental uncertainties. While reinforcement learning (RL) has emerged as a promising alternative to traditional model-based control, existing approaches frequently rely on flat training pa