Machine learning and evolutionary techniques in interplanetary trajectory design

TL;DR

This paper explores the integration of machine learning and evolutionary algorithms to enhance interplanetary trajectory design, demonstrating the potential of neural networks for onboard guidance in space missions.

Contribution

It introduces the novel application of deep neural networks for onboard guidance profile representation in interplanetary missions, expanding previous research to orbital transfers.

Findings

Neural networks can effectively represent optimal guidance profiles.

Application demonstrated on Earth-Mars transfer case.

Extends previous landing and quadcopter dynamics research.

Abstract

After providing a brief historical overview on the synergies between artificial intelligence research, in the areas of evolutionary computations and machine learning, and the optimal design of interplanetary trajectories, we propose and study the use of deep artificial neural networks to represent, on-board, the optimal guidance profile of an interplanetary mission. The results, limited to the chosen test case of an Earth-Mars orbital transfer, extend the findings made previously for landing scenarios and quadcopter dynamics, opening a new research area in interplanetary trajectory planning.