Nebulae: A Proposed Concept of Operation for Deep Space Computing Clouds

In this paper, we describe an ongoing multi-institution study in using emplaced computational resources such as high-volume storage and fast processing to enable instruments to gather and store much more data than would normally be possible, even if it cannot be downlinked to Earth in any reasonable time. The primary focus of the study is designing science pipelines for on-site summarization, archival for future downlink, and multisensor fusion. A secondary focus is on providing support for increasingly autonomous systems, including mapping, planning, and multi-platform collaboration. Key to both of these concepts is treating the spacecraft not as an autonomous agent but as an interactive batch processor, which allows us to avoid “quantum leaps” in machine intelligence required to realize the concepts. Our goal is to discuss preliminary results and technical directions for the community, and identify promising new opportunities for multi-sensor fusion with the help of planetary researchers.