Title:
A Technical Debt Management Framework for Aerospace Systems Engineering: An AI-Driven Approach to Test and Evaluation Documentation Analysis
Abstract:
Unmanaged technical debt in aerospace systems engineering has produced consequences ranging from the Hubble Space Telescope's $629 million on-orbit correction to the Mars Climate Orbiter's $327.6 million total mission loss. Despite three decades of growing research attention, detection methods remain anchored in software-centric assumptions that do not hold at the physical integration boundaries where aerospace systems are most vulnerable. This dissertation develops and validates a Technical Debt Management Framework (TDMF) for identifying, classifying, and prioritizing technical debt during the test and evaluation (T&E) phase of systems engineering projects. The framework is operationalized through the Aerospace Technical-debt Learning and Assessment System (ATLAS), a domain-specific AI tool that applies large language model classification to unstructured T&E documentation.
The research follows a Design Science Research methodology with three evaluation streams: ATLAS classifier performance on 141 aerospace T&E documents (F1 = 0.82, κ = 0.84), retrospective validation against the HST and MCO programs, and an expert survey (n = 35) confirming 97% positive adoption intent and a 45% reduction in median review time.
This work contributes to the body of knowledge by:
- Introducing a systems-engineering-specific technical debt taxonomy with bidirectional leading-indicator mappings calibrated to aerospace practice.
- Validating an AI architecture demonstrating that few-shot LLM classification achieves documentation-scale analysis in safety-critical domains without task-specific fine-tuning.
- Providing empirical evidence that the framework resolves the asymmetric detection gap at hardware–software boundaries where 78% of aerospace technical debt manifests.
Department of Systems Engineering | Worcester Polytechnic Institute 100 Institute Road | Worcester, MA 01609-2280
Advisor:
Prof. Shamsnaz Bhada
ECE Department, WPI
Committee Members:
Prof. Jamie Monat
ECE Department, WPI
Beth Wilson
ECE Department, WPI
Larry Rosser
RTX Corporation