Close-Proximity Satellite Operations through Deep Reinforcement Learning and Terrestrial Testing Environments

TL;DR

This paper investigates the use of deep reinforcement learning for autonomous satellite control in congested space, testing algorithms from simulation to real-world hardware to assess robustness and performance degradation.

Contribution

It introduces DRL-based control methods for satellite operations and evaluates their effectiveness across simulated and real terrestrial testing environments.

Findings

DRL algorithms perform well in simulations

Performance degrades when transferred to real hardware

Environmental disturbances impact control accuracy

Abstract

With the increasingly congested and contested space environment, safe and effective satellite operation has become increasingly challenging. As a result, there is growing interest in autonomous satellite capabilities, with common machine learning techniques gaining attention for their potential to address complex decision-making in the space domain. However, the "black-box" nature of many of these methods results in difficulty understanding the model's input/output relationship and more specifically its sensitivity to environmental disturbances, sensor noise, and control intervention. This paper explores the use of Deep Reinforcement Learning (DRL) for satellite control in multi-agent inspection tasks. The Local Intelligent Network of Collaborative Satellites (LINCS) Lab is used to test the performance of these control algorithms across different environments, from simulations to real-world quadrotor UAV hardware, with a particular focus on understanding their behavior and potential degradation in performance when deployed beyond the training environment.