Quantum Reinforcement Learning for Joint Control, Communication, and Computing in Stabilized Reusable Space Rocket

초록

Reusable rockets have emerged as a pivotal technology for cost reduction and efficiency improvement in the space industry. Joint control, communication, and computing technologies must be considered for the mission performance of these reusable rockets. This article introduces a quantum reinforcement learning (RL) for the joint control, communication, and computing (QRL-C3) algorithm designed for quantum reusable rockets. The QRL-C3 aims to solve the issue of diminished training performance in high-dimensional states and action spaces prevalent in conventional RL. Leveraging quantum neural networks with superposition and entanglement, QRL-C3 enhances information representation and processing. The framework integrates a 1) robust quantum control system for precise reentry and landing; 2) a quantum multiagent RL-based scheduler for seamless communication with multiple ground stations for enabling Internet-of-Things services; and 3) energy-efficient computing algorithms considering queue backlog for optimal image processing and transmission power management. These components synergistically improve mission performance and rocket reusability. Experimental validation using SpaceX's Falcon 9 model demonstrates that QRL-C3 outperforms conventional RL approaches, achieving higher training efficiency. This study is the first to integrate control, communication, and computing systems in reusable rockets through quantum artificial intelligence (QAI), paving the way for future advancements in QAI-based space technologies.

키워드