Deep Reinforcement Learning Based Systems for Safety Critical Applications in Aerospace

TL;DR

This paper discusses the integration of deep reinforcement learning into aerospace control systems, highlighting its potential to enhance real-time monitoring, fault detection, and autonomous operation on advanced HPC platforms.

Contribution

It introduces the application of deep reinforcement learning to aerospace control systems, emphasizing its role in safety-critical, real-time, autonomous, and assistive functions.

Findings

Deep reinforcement learning can improve fault detection in aerospace systems.

HPC platforms enable real-time AI applications in aerospace control.

AI can augment human control with autonomous features.

Abstract

Recent advancements in artificial intelligence (AI) applications within aerospace have demonstrated substantial growth, particularly in the context of control systems. As High Performance Computing (HPC) platforms continue to evolve, they are expected to replace current flight control or engine control computers, enabling increased computational capabilities. This shift will allow real-time AI applications, such as image processing and defect detection, to be seamlessly integrated into monitoring systems, providing real-time awareness and enhanced fault detection and accommodation. Furthermore, AI's potential in aerospace extends to control systems, where its application can range from full autonomy to enhancing human control through assistive features. AI, particularly deep reinforcement learning (DRL), can offer significant improvements in control systems, whether for autonomous operation or as an augmentative tool.