Reinforcement Learning for Robust Trajectory Design of Interplanetary Missions

What is it about?

This paper investigates the use of machine learning for the design of low-thrust interplanetary trajectories that are robust to several sources of uncertainties and disturbances. In particular, we consider alternatively errors on the dynamical model, observation noise, control actuation errors, and missed thrust events. A state-of-the-art algorithm, named Proximal Policy Optimization, is adopted to carry out the optimization procedure. Numerical results are presented for an Earth-Mars mission to demonstrate the robustness and effectiveness of the proposed approach.

Featured Image