Initial Guess Generation for Low-Thrust Trajectory Design with Robustness to Missed-Thrust-Events
Amlan Sinha, Ryne Beeson
TL;DR
This paper addresses robust low-thrust trajectory design under missed-thrust-event (MTE) uncertainty in cislunar missions. It introduces and compares two global-search initial-guess strategies: a non-conditional approach that samples broadly from a fixed distribution and a conditional approach that uses solutions from simpler, less robust problems to guide the search. The authors formalize a robust LT trajectory framework with two formulations—the General MTE Formulation (allowing multiple realizations) and the Restricted Finite Realization MTE Formulation (limiting to a finite number of MTEs) and provide a nonlinear-program transcription for practical optimization. Through a case study inspired by the Lunar Gateway Power and Propulsion Element, they demonstrate that the conditional initialization markedly improves convergence rate and solution quality, enabling faster and more reliable exploration of robust LT transfers in a high-dimensional design space. Overall, the work offers a practical, statistically validated approach to incorporating robustness into preliminary mission design, with significant implications for efficiently planning LT trajectories subject to MTEs.
Abstract
The growing interest in cislunar space exploration in recent years has driven an increasing demand for efficient low-thrust missions to key cislunar orbits. These missions, typically possessing long thrust arcs, are particularly susceptible to operational uncertainties such as missed thrust events. Addressing these challenges requires efficient robust trajectory design frameworks during the preliminary mission design phase, where it is necessary to explore the solution space at a rapid cadence under evolving operational constraints. However, existing methods for missed thrust design rely on solving high-dimensional nonlinear programs, where generating effective initial guesses becomes challenging. To enhance computational efficiency, quality, and depth of robustness of solutions from global search, we compare two initial guess strategies: a baseline non-conditional global search, which samples from a static distribution with global support, and a conditional global search, which generates initial guesses conditioned on solutions to problems with less depth of robustness. The conditional search provides a sequential procedure for solving increasingly robust problems. We validate the improvements in the conditional approach using a low-thrust case study for the Lunar Gateway Power and Propulsion Element, where our results demonstrate that it significantly improves convergence rate and solution quality, highlighting its potential in preliminary robust trajectory design.