TraceAI
Develop a new watermarking/blockchain technological approach that hardens against adversarial Al while ensuring the provenance, integrity, and quality of space-based artificial intelligence algorithms and training data for CubeSats and high-throughput space infrastructure.
Technical Merit
- Improve cybersecurity in space to enable a larger scale, higher throughput, and a more securely interconnected ecosystem.
- Enable on-orbit manufacturing capabilities to supply large scale space production industries.
- Develop a process for monitoring secure and remote updates of AI/ML Models
Key Participants
Project Lead Agency
National Science Foundation
Academic Partners
UTSA
Commercial Partner
Constellation Network
Technology
- Al/ML model governance process designed to harden against adversarial Al attacks and threats while delivering improved development efficiency
- Scalable DAG blockchain designed, deployed, and optimized to maintain the pedigree and immutability of Al/ML models and associated training data
- Dual use Al model development documentation process that enables immutable tracing and verification of intellectual property rights and copyright clearances
- Tracing and tracking of AI.ML model training data to support VR/AR Simulation Balance testing requirements
- CI/CD pipeline designed to enable autonomous on-orbit updates of Deep Learning algorithms.
