Deep Learning-Based Robust Optical Guidance for Hypersonic Platforms

TL;DR

This paper introduces a deep learning approach for robust optical guidance of hypersonic platforms by encoding image stacks into a neural network, addressing limitations of classical registration methods especially in bimodal scenes.

Contribution

The paper proposes a novel deep learning method that encodes multiple images into a network to improve guidance accuracy in challenging conditions.

Findings

Effective in bimodal scenes such as snowy and non-snowy environments

Outperforms classical registration methods in robustness

Demonstrates potential for long-range hypersonic guidance

Abstract

Sensor-based guidance is required for long-range platforms. To bypass the structural limitation of classical registration on reference image framework, we offer in this paper to encode a stack of images of the scene into a deep network. Relying on a stack is showed to be relevant on bimodal scene (e.g. when the scene can or can not be snowy).