Epscor research fellows: NSF: AI-secure command channels for autonomous space systems

The National Science Foundation Office of Integrative Activities awarded New Mexico State University $299,665 under the Integrative Activities program (CFDA 47.083) effective January 1, 2027, through December 31, 2028. The award funds an EPSCOR Research Fellows project at New Mexico State University's Las Cruces campus that provides a fellowship to an assistant professor and training for one student trainee. The research investigates how artificial intelligence can be used safely to support command decisions for autonomous spacecraft, with focus on detecting and preventing cyberattacks that manipulate AI-generated spacecraft commands before they affect mission operations. The project will develop a formal framework for modeling how AI systems transform inputs into spacecraft commands, characterize AI-enabled command manipulation attacks, and develop runtime mechanisms that validate AI-generated commands against mission and safety constraints before execution. Methodology includes formal threat modeling, taxonomy development, experimental prototype implementation, and empirical evaluation using an AI-enabled satellite command prototype combining AI reasoning components with realistic spacecraft command and telemetry interfaces. The University of California, Santa Cruz serves as a collaborator. The fellowship strengthens research capacity in New Mexico by training the next generation of cybersecurity researchers and expanding institutional collaboration while advancing technologies that improve the reliability, safety, and security of future space missions.

Abstract

This research infrastructure improvement (RII) epscor research fellows project provides a fellowship to an assistant professor and training for one student trainee at new mexico state university. This work will be conducted in collaboration with collaborators at the university of california, santa cruz. Through the fellowship, the PI will investigate how artificial intelligence (AI) can be used safely to support command decisions for autonomous spacecraft. The project will develop methods to detect and prevent cyberattacks that manipulate AI?generated spacecraft commands before they can affect mission operations. This research combines cybersecurity, artificial intelligence, and aerospace engineering to improve the security and resilience of future autonomous space systems. The fellowship will strengthen research capacity in new mexico by training the next generation of cybersecurity researchers, expanding collaboration between the two institutions, and advancing technologies that improve the reliability, safety, and security of future space missions. This project will establish a research program focused on securing AI-enabled spacecraft. The project will advance the state of the art by introducing a formal framework for modeling how AI systems transform inputs into spacecraft commands, characterizing AI-enabled command manipulation attacks, and developing runtime mechanisms that validate AI-generated commands against mission and safety constraints before execution. The proposed methodology will integrate formal threat modeling, taxonomy development, experimental prototype implementation, and empirical evaluation using an AI-enabled satellite command prototype that combines AI reasoning components with realistic spacecraft command and telemetry interfaces. The fellowship will strengthen research infrastructure at new mexico state university by expanding faculty expertise in AI-enabled cyber-physical systems, establishing new collaborations with the university of california, santa cruz, and providing research training for undergraduate and graduate students. The fellowship will support the development of a sustainable research program, future external funding, curriculum development in AI and cybersecurity, and a workforce prepared to secure next-generation space systems. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the foundation's intellectual merit and broader impacts review criteria.- subawards are not planned for this award.