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

2025-02-27

AI Summary

To address safety and generalization challenges in multi-satellite cooperative rendezvous under highly dynamic, interference-prone low Earth orbit (LEO) conditions, this work proposes a cross-domain verifiable deep reinforcement learning (DRL) framework for space operations. Methodologically, it integrates proximal policy optimization (PPO) and soft actor-critic (SAC) algorithms with multi-agent cooperative control, and establishes a three-tier validation platform: high-fidelity simulation, hardware-in-the-loop testing, and real-world deployment on quadcopters (LINCS Lab). We introduce a novel perturbation robustness analysis paradigm to systematically characterize performance degradation under sensor noise, environmental disturbances, and control latency. Experiments demonstrate centimeter-level relative positioning accuracy, sub-second response latency, and >86% cross-domain deployment success—significantly outperforming PID/LQR baselines. The framework further enhances decision interpretability and trustworthiness, establishing a verifiable and transferable paradigm for onboard autonomous spacecraft control.

Technology Category

• Intelligent Robots: State Estimation
• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments

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.

Problem

• Autonomous satellite operations in congested space environments
• Understanding DRL model sensitivity to environmental disturbances
• Testing DRL algorithms in simulations and real-world UAV hardware

Innovation

• Deep Reinforcement Learning for satellite control
• Testing across simulation and real-world environments
• Focus on performance degradation beyond training