Terrestrial high-performance computing has enabled incredible advances in autonomous vehicles, machine learning, and data analysis for Earth-side applications. MOSAIC seeks to bring together JPL’s High-Performance Spaceflight Computing (HPSC) capabilities, innovative onboard autonomy, and novel mission concepts based on these technologies to create comprehensive guidance documents, trade studies, and functional, scalable flight hardware ready for mission use. Increasing numbers of assets on Mars, each with increasing CPU capabilities, creates unique opportunities to use distributed (cloud-like), self-organizing and networked computing for future mission concepts.
MOSAIC will build upon proven Deep Space Network (DSN) technologies such as delay-tolerant-networking (DTN) combined with HPSC to build an ever-improving network of shared computing, storage, and optimized data routing between stationary, roving, and orbital assets. Assets on a MOSAIC network may divide work by requesting or providing data sources, computation, storage, or network relay. This task focuses on three objectives:
1) Build software tool for trade space exploration that can optimize locations of assets, their communication requirements, roles, and software capabilities to maximize mission impact.
2) Build software library to enable distributed computation over ad-hoc networked devices and to enable a cloud-like service oriented distributed architecture in future missions.
3) Implement and demonstrate validated software incorporating DTN networking on representative CPU architectures, rovers, and with realistic mission use-cases.
Joshua Vander Hook - Jet Propulsion Laboratory
Research and Technology Development Program