System Framework for Autonomous Data Processing Onboard Next Generation of Nanosatellite

Progress within nanosatellite systems development makes niche commercial Earth observing missions feasible; however, despite advances in demonstrated data rates, these systems will remain downlink limited able to capture more data than can be returned to the ground cost-effectively in traditional raw or near-raw forms. The embedding of existing ground-based image processing algorithms into onboard systems is non-trivial especially in limited resource nanosatellites, necessitating new approaches. In addition, mission opportunities for systems beyond Earth orbit present additional challenges around relay availability and bandwidth, and delay-tolerance, leading to more autonomous approaches. This paper describes a framework for implementing autonomous data processing onboard resource-constrained nanosatellites, covering data selection, reduction, prioritization and distribution. The framework is based on high level requirements and aligned to existing off-the-shelf software and international standards. It is intended to target low-resource algorithms developed in other sectors including autonomous vehicles and commercial machine learning. Techniques such as deep learning and heuristic code optimization have been identified as both value-adding to the use cases studied and technically feasible. With the framework in place, work is now progressing within the consortium under UKSA Centre for Earth Observation and Instrument funding to deliver an initial prototype data chain implemented within a representative FPGA-based flight computer system.