Secure Autonomous Computing Stack

Objective: Deliver verified AI assurance methodologies that improve autonomous system quality and provide the confidence evidence required for mission-critical deployment of ASTRA capabilities. This initiative addresses the fundamental challenge that autonomous spacecraft operating in contested, delayed, and denied communication environments cannot depend on ground intervention when AI-driven decisions determine mission success. The proposed AI Assurance Framework provides the verification and validation foundation that makes trusted autonomy possible in safety-critical space applications. By establishing rigorous V&V methodologies specifically designed for AI's non-deterministic behaviors, we transform autonomous capabilities from promising technology demonstrations into mission-qualified systems. The proposed innovation develops a holistic assurance framework extending beyond traditional software assurance to encompass the entire AI system lifecycle. This includes robust requirements engineering for AI-based systems, specialized design assurance for non-deterministic behaviors, rigorous data governance protocols, continuous monitoring techniques, and AI-specific risk management strategies. To systematically improve AI system quality and establish mission confidence, we address critical research questions: What supplementary approaches beyond established V&V techniques guarantee consistent AI behaviors? How do we ensure continuous collection of accurate, complete data for AI monitoring and prediction? Are traditional safety envelopes sufficient, or do AI systems require alternative safety designs? How can Explainable AI and white-box analysis integrate into early development for formal verification? What methodologies validate "Double-AI" approaches where AI techniques verify AI systems? Direct Support for ASTRA Technology Pathways: This framework enables trust and certification for Trusted Autonomy and Resilience (verification for anomaly detection and self-healing), Autonomous Fault Detection, Isolation, and Recovery (safety envelopes for AI-driven fault response), and AI/ML-driven onboard science decision-making (explainable autonomous observation prioritization). This work addresses the confidence gap limiting AI infusion: stakeholders recognize autonomous capabilities' potential but lack assurance evidence for operational approval. We will continue coordinating with Code 700's CAIO to ensure institutional alignment, with IV&V to provide independent validation expertise for safety-critical operations, and with LaRC, ARC, and JPL research groups to leverage cross-center expertise in autonomous systems verification. Through rigorous synthesis of academic research, state of practice from other centers and agencies, and industry best practices, we will deliver a practical AI Assurance Framework addressing the unique challenges of opacity, emergent behaviors, and evolving capabilities in NASA's autonomous flight missions.

Holistic AI Assurance Framework for verification and validation (V&V) of autonomous spacecraft AI systems throughout the full lifecycle; rigorous V&V methodologies for non-deterministic AI behaviors; robust requirements engineering, design assurance, data governance, continuous monitoring, and AI-specific risk management; enables Explainable AI and white-box analysis integration; "Double-AI" validation methodologies; direct enablement of ASTRA Trusted Autonomy, Autonomous Fault Detection/Isolation/Recovery, and AI/ML-driven onboard science decision-making capabilities.

This is a mature technology (TRL 7+) — the realistic path in is usually NASA's Technology Transfer Program: licensing an existing NASA patent, or a Space Act Agreement to use NASA facilities/expertise directly. NASA also runs a startup licensing program with no upfront fee for companies formed to commercialize a specific NASA technology.

None of these are guaranteed paths for this specific project — TechPort itself doesn't have an "apply" button. Reaching out to the contact(s) above with a specific question is usually the fastest way to find out what's actually open.