Verifiable Success in handling Unknown Unknowns in Space Habitat Simulations and a Cyber-Physical System

We propose a layered approach which is, adaptive to different levels of computational resources: 1. Formal Methods – Prof. Kristin Yvonne Rozier expands upon NASA’s R2U2 framework, integrating proactive strategies for handling “unknown unknowns” by embedding learning and exploration directly into formal verification methods ensuing compliance with NASA’s verifiability requirements. 2. Physics-Based Bayesian reasoning – Prof. Ilias Bilionis takes the next step with the goal to improve resilience and computational efficiency, strategically targeting challenges presented by unforeseen events (“unknown unknowns”). The models are currently one of the practical approaches for handling a dynamic environment with unforeseeable changes. enabling dynamic adaptation autonomous space habitat operations. 3. Large Language Model (LLM)-Driven Deep Reasoning – Hedinn Steingrimsson and his team tailor cutting-edge deep reasoning architectures utilizing LLMs to bolster system resilience, especially for root-cause analyses involving unknown-unknown scenarios. In close collaboration with Prof. Rozier, we leverage the strength of recent advances in neural reasoning while integrating formal verification constraints and where formal models provide safety guarantees. 1. Cyber-Physical Space Habitat Integration – Prof. Shirley Dyke applies and adapts our multidisciplinary and multimodal approach. We rigorously evaluate the practical applicability of our models in real world space habitat. With realistic experimental conditions we to rigorously evaluate our models for both fully autonomous and also crew on board scenarios. Funding directly supports theoretical advancements, experimental validations aimed at fundamentally improving “unknown unknowns” capabilities in a manner that can be fully integrated with NASA’s current operational systems. With subsequent domain adaptation valuable assets can be created for performance in dynamic operational environments across different domains.

Lunar Gateway Autonomy – Current TRL: 4-5. Key components of our approach are already at mid-TRL in the context of Gateway – for instance, the R2U2 runtime monitor which we improve and give capabilities to handle “unknown unknowns” is currently operating in Gateway’s test environment, indicating about TRL 5 (validated in relevant environment). Our Bayesian and causal reasoning is at TRL 3-4; it’s been proven in principle (e.g. causal fault models in NPAS). Artemis Base Camp – Autonomous Lunar Surface Habitat. Will include long time periods without onsite crew. Our technology will enable the Base Camp to autonomously manage power, life support, termal, food system and or payload as well as integrity in a harsh environment (deep cold, lunar dust) to remain operational. Moon-to-Mars – e.g. Mars Campaign Development, Mars Sample Return, Dragonfly and crewed Mars missions starting with the Gateway serving as the first space habitat and Future Orbital Habitats. Fundamentally Mars requires the technology that we develop both due to a time period without any communication to an earth based control center when Mars is on the other side of the sun and also due to long – up to 40 minutes round trip - communication latency and low communication bandwith. Thus fully autonomous systems are necessary including for in orbit as well as ground based space habitats and stations. Unlike the Moon, an immediate return or rescue is impossible, demanding ultra-reliable autonomy for crew safety. Our deliverable is designed to provide maximum possible resilience under extreme conditions over a long period of time, where material wear and tear is to be expected and the system must operate without external aid. Deep Space Transport, built on Gateway autonomy and habitat autonomony. Our deliverables can eventually contribute to the Game Changing Development (GCD) Autonomous building management systems: private and public sector buildings including archeological museum and libraries which need to maintain certain operational parameters to protect both people and also assets e.g. archeological items. There are recent examples e.g. in the California wildfires, where capable systems together with prophylaxis anti-fire measures managed to protect buildings. Our solution improves the capability of the systems to successfully handle adverse scenarios. Nuclear and chemical plant management: are safety critical system with in many ways a similar operational environment, with necessary formal verification, resilience and reliability. AI has entered this domain in a limited capacity. We envision that our system could be adapted to maintaining the integrity of safety critical systems where the risk comes from within rather than externally. The same fundamental principles apply as well as similar operational goal. We envision that leveraging our technology at early stages in this domain builds valuable knowledge and experience which would create synergies since the reliability of our system would not only be evaluated in the space domain. Thus we envision that a thrust here will be very important in improving the reliability and resilience of our technology through extensive testing in critical operational environments. Air traffic control: currently AI has in a limited capacity entered air traffic control, with the situation in many ways being similar as at NASA with strict safety regulations. Our system runs both fully autonomously and also with a human in the loop. The Purdue Space habitat enables us to test both settings. We envision that our system could be adapted to air traffic control and other safety-critical operative environments. The domain transfer requires extensive testing: run our system in parallel with the current operational environment and when experience and trust has been accomplished gradually playing larger role.

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.