Search

The orbits around earth are a limited natural resource and one that hosts a vast range of vital space-based systems that support international systems use by both commercial industries, civil organisations, and national defence. The availability of this space resource is rapidly depleting due to the ever-growing presence of space debris and rampant overcrowding, especially in the limited and highly desirable slots in geosynchronous orbit. The field of Space Situational Awareness encompasses tasks aimed at mitigating these hazards to on-orbit systems through the monitoring of satellite traffic. Essential to this task is the collection of accurate and timely observation data. This thesis explores the use of a novel sensor paradigm to optically collect and process sensor data to enhance and improve space situational awareness tasks. Solving this issue is critical to ensure that we can continue to utilise the space environment in a sustainable way. However, these tasks pose significant engineering challenges that involve the detection and characterisation of faint, highly distant, and high-speed targets. Recent advances in neuromorphic engineering have led to the availability of high-quality neuromorphic event-based cameras that provide a promising alternative to the conventional cameras used in space imaging. These cameras offer the potential to improve the capabilities of existing space tracking systems and have been shown to detect and track satellites or ‘Resident Space Objects’ at low data rates, high temporal resolutions, and in conditions typically unsuitable for conventional optical cameras. This thesis presents a thorough exploration of neuromorphic event-based cameras for space situational awareness tasks and establishes a rigorous foundation for event-based space imaging. The work conducted in this project demonstrates how to enable event-based space imaging systems that serve the goals of space situational awareness by providing accurate and timely information on the space domain. By developing and implementing event-based processing techniques, the asynchronous operation, high temporal resolution, and dynamic range of these novel sensors are leveraged to provide low latency target acquisition and rapid reaction to challenging satellite tracking scenarios. The algorithms and experiments developed in this thesis successfully study the properties and trade-offs of event-based space imaging and provide comparisons with traditional observing methods and conventional frame-based sensors. The outcomes of this thesis demonstrate the viability of event-based cameras for use in tracking and space imaging tasks and therefore contribute to the growing efforts of the international space situational awareness community and the development of the event-based technology in astronomy and space science applications.

In this thesis, I demonstrate a novel and efficient unsupervised clustering and data exploration method with the combination of a Self-Organising Map (SOM) and a Convolutional Autoencoder, applied to radio-astronomical images from the Radio Galaxy Zoo (RGZ) dataset. The rapidly increasing volume and complexity of radio-astronomical data have ushered in a new era of big-data astronomy which has increased the demand for Machine Learning (ML) solutions. In this era, the sheer amount of image data produced with modern instruments and has resulted in a significant data deluge. Furthermore, the morphologies of objects captured in these radio-astronomical images are highly complex and challenging to classify conclusively due to their intricate and indiscrete nature. Additionally, major radio-astronomical discoveries are unplanned and found in the unexpected, making unsupervised ML highly desirable by operating with few assumptions and without labelled training data. In this thesis, I developed a novel unsupervised ML approach as a practical solution to these astronomy challenges. Using this system, I demonstrated the use of convolutional autoencoders and SOM’s as a dimensionality reduction method to delineate the complexity and volume of astronomical data. My optimised system shows that the coupling of these methods is a powerful method of data exploration and unsupervised clustering of radio-astronomical images. The results of this thesis show this approach is capable of accurately separating features by complexity on a SOM manifold and unified distance matrix with neighbourhood similarity and hierarchical clustering of the mapped astronomical features. This method provides an effective means to explore the high-level topological relationships of image features and morphology in large datasets automatically with minimal processing time and computational resources. I achieved these capabilities with a new and innovative method of SOM training using the autoencoder compressed latent feature vector representations of radio-astronomical data, rather than raw images. Using this system, I successfully investigated SOM affine transformation invariance and analysed the true nature of rotational effects on this manifold using autoencoder random rotation training augmentations. Throughout this thesis, I present my method as a powerful new approach to data exploration technique and contribution to the field. The speed and effectiveness of this method indicates excellent scalability and holds implications for use on large future surveys, large-scale instruments such as the Square Kilometre Array and in other big-data and complexity analysis applications.