Too Distant, Too Uncertain, Too Unforgiving: Assuring Autonomous Space Explorers We’ve Yet to Trust

Symposium Overview:

The primary goal of this symposium is to initiate a collaborative investigation into a new assurance paradigm and to catalyze a community capable of realizing it. Our specific objectives include:

• Establish a Multi-Disciplinary Community: Convene a currently sparse group of experts from JPL, Caltech, and the broader aerospace, automotive, and defense sectors to foster long-term collaborations.
• Identify Gaps: Explore high-stakes use cases to pinpoint where traditional ground-in-the-loop and testing-based V&V currently fail. Collaboratively define the boundaries of existing assurance practices and standards when applied to adaptive, non-deterministic space systems.
• Investigate “Assurance as a System Function”: Conduct an initial investigation into the architectural abstractions and formal models required for runtime monitoring and self-governance. Address the need for a principled process to assure the very algorithms that monitor, assess, and evolve the system’s internal models. Identify high-priority gaps in tools, algorithms, and auditable processes to guide future technology investments and joint research proposals.
• Empower Early-Career Researchers: Provide a platform for graduate students and postdocs to present recent discoveries in formal methods and safe learning, integrating their research into the broader mission-assurance conversation.