AIRMATE: AIRworthiness Model Automation for Certification Evaluations

To address the United States Air Force’s need for efficient and cost-effective transformation of certification documents to MBSE-ready digital models, RedShred proposes to continue developing the AIRworthiness Model Automation for cerTification Evaluations system (AIRMATE) based on the latest advances in generative AI and Redshred’s existing document AI platform. Specifically, the innovation in automatically extracting digital certification artifacts for digital models from legacy documents will enable the technology to accelerate certification processes, eliminate the need to manually search across multiple documents, increase transparency, reduce cost and labor, and empower downstream tasks that depend on the successful parsing of the certification documents (e.g., modification description alignment with relevant criteria). As a result, this system offers a scalable, lightweight technology and does not require large datasets to support domain adaptability. In addition, AIRMATE supports multiple formats, which directly addresses the customer requirements to be compatible with popular third-party MBSE tools. Redshred will fulfill all Phase II requirements established by relevant stakeholders and demonstrate the feasibility of using advanced generative AI to transform certification documents to MBSE-ready digital models in multiple domains with cross-reference linking. In Phase II, Redshred will 1) Review the System Architecture and Requirements, 2) Acquire & Process Documents, Update SysML Mapping, 3) Prompt Refinement & Development of Quality Evaluation Models 4) Cross-Document Linking & Extraction, 5) Modification Specification & Criteria Alignment Pipeline, 6) UI Development for Easy User Interaction, and 7) Develop SysML Outputs and Testing of Feasibility.