A survey on artificial intelligence trends in spacecraft guidance dynamics and control

Abstract

The rapid developments of Artificial Intelligence in the last decade are influencing Aerospace Engineering to a great extent and research in this context is proliferating. We share our observations on the recent developments in the area of Spacecraft Guidance Dynamics and Control, giving selected examples on success stories that have been motivated by mission designs. Our focus is on evolutionary optimisation, tree searches and machine learning, including deep learning and reinforcement learning as the key technologies and drivers for current and future research in the field. From a high-level perspective, we survey various scenarios for which these approaches have been successfully applied or are under strong scientific investigation. Whenever possible, we highlight the relations and synergies that can be obtained by combining different techniques and projects towards future domains for which newly emerging artificial intelligence techniques are expected to become game changers.

Low-Thrust Trajectory Design Using Closed-Loop Feedback-Driven Control Laws and State-Dependent Parameters

“…For a spacecraft, this could arise from orbit determination uncertainties or thruster misalignment. There is significant interest in the applications of RL for mission design, operations, guidance and control to navigation and even the prediction of the dynamics [20]. However, little work has combined RL with existing control laws for trajectory design, which would result in powerful state-dependent learning process on top of an intelligent controller.…”

“…One method of achieving this is using evolutionary algorithms (EAs). The multi-objective nature of spacecraft trajectory design lends itself to EAs and they still provide many of the benchmarks in the area [20]. Both Lee et al 2005 [8] and Varga et al 2016 [9] used a multi-objective genetic algorithm to optimise the Q-law design parameters for a variety of Earth orbit transfers, with the design parameters remaining fixed throughout the transfer.…”

“…Reinforcement learning is a promising alternative increasingly used in astrodynamics [20]. Gaudet, Furfaro and Linares have demonstrated the capability of RL for use in Mars landers [22] and asteroid hovering missions [23].…”

A Pre-Trained Fuzzy Reinforcement Learning Method for the Pursuing Satellite in a One-to-One Game in Space

“…Tracking space targets is beneficial for orbital garbage removal, recovery of important components, and early warning of space threats [1]. However, with the continuous development of space techniques, the targets in space have been expanded from non-maneuverable ones to maneuverable ones.…”

Interplanetary transfers via deep representations of the optimal policy and/or of the value function

“…The use of deep neural networks (DNNs) for the guidance navigation and control of space systems (spacecraft, landers, etc..) is a prolific area of research as witnessed by the increasing number of results that appeared recently on these topics [9,11,3,12,4,6,14]. Most of this body of work can be divided into DNNs approximating the optimal policy u (e.g.…”