Back to search

University of Strathclyde

Robust artificial intelligence for space traffic management

Abstract

dc:description.abstract

This work proposes a new methodology to manage the increase in space traffic and ensure the safe and sustainable utilisation of space by all actors involved in the New Space context. The proposed methodology aims to improve the automation and robustness of Space Traffic Management. The resulting framework constitutes CASSANDRA, an intelligence agent to robustly support operators on complex Space Traffic Management tasks accounting for aleatory and epistemic uncertainty and implementing Artificial Intelligence techniques to address the problems.This research presents an evidence-based framework to assist operators in the conjunction risk assessment decision-making process, using Dempster-Shafer theory of evidence to model both aleatory and epistemic uncertainty on the object’s state vector. This framework assists operators in making robust decisions in space conjunction assessment based on the value of the probability of collision and its correctness from the available information. The framework is designed to cope with Conjunction Data Messages. These messages are the most common standard protocol for conjunction communication, and the proposed methodology models the epistemic uncertainty affecting them. The framework also addresses the Conjunction Avoidance Manoeuvre design by providing robust optimal strategies accounting for both aleatory and epistemic uncertainty.The framework enhances the autonomy of Space Traffic Management using Artificial Intelligence methods. These techniques facilitate the autonomy of the decision processes through the creation of faster surrogate models, data-driven models and decision-making architectures. The use of Artificial Intelligence in this research intends to improve the automation of the Space Traffic Management system. First, a Decision Support System based on Multi-Criteria Decision-Making and Game Theory is proposed to prioritise the best avoidance strategies based on the available alternatives and the operator’s criteria and constraints. Second, automation is improved by implementing Machine Learning and Deep Learning techniques, like Random Forests or Neural Networks, to speed up the conjunction assessment process accounting for both types of uncertainty while providing reliable levels of accuracy.Finally, this work presents some examples where the methodology is tested on a rangeof real and synthetic scenarios addressing the multiple-encounter events problem, presenting a pipeline integrating the different elements of the framework, and comparing the proposed framework with the current approaches followed by the European and French Space Agencies.

Degree

thesis:*
Name dc:type.qualificationname
phd
Level dc:type.qualificationlevel
doctoral-pg
Grantor dc:publisher.institution
University of Strathclyde
Year dc:date.issued
2025

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Fernández-Mellado, Luis Sánchez
Advisor dc:contributor.advisor
  • Vasile, Massimiliano
Author Identifier dc:creator.authoridentifier
  • 201873916

Rights

Identifiers

dc:identifier.*
Identifier
T17306

Chain of custody

source
Harvested from
University of Strathclyde
Base URL
stax.strath.ac.uk/catalog/oai
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
related terms
citation

Fernández-Mellado, Luis Sánchez. Robust artificial intelligence for space traffic management. doctoral-pg thesis, University of Strathclyde, 2025. https://stax.strath.ac.uk/concern/theses/w9505114f