New reinforcement learning method optimizes interplanetary transfers

Researchers have developed a novel approach called Reachability Analysis-Informed Reinforcement Learning (RARL) to design interplanetary spacecraft trajectories. This method integrates reinforcement learning with reachability maps to select intermediate waypoints, which then inform classical astrodynamics techniques for maneuver planning. In tests on an Earth-Mars benchmark, RARL achieved a mean maneuver cost close to that of sequential convex programming and demonstrated significant policy reuse across various departure conditions, with independently trained policies successfully completing all test departures without violating impulse constraints. AI

IMPACT This research could lead to more efficient and reusable automated systems for space mission planning.

RANK_REASON Academic paper detailing a new methodology for spacecraft trajectory design. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• alphaXiv
• arXiv
• CatalyzeX
• DagsHub
• Earth
• Gotit.pub
• Hugging Face
• Lambert reconstruction
• Mars
• Reachability Analysis-Informed Reinforcement Learning
• reinforcement learning
• Sequential Convex Programming
• Yashdeep Chaudhary

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →