Autonomy Levels for Small Satellite Clusters

The growth in commercial and government use of small satellites and CubeSat constellations is projected to grow significantly over the next decade due to their low cost and flexibility attributes. However, managing clusters of small satellites requires some level of distribution of cognition amongst the elements of the cluster to reduce human workload and enable autonomous and robust operations. The goal of introducing cognition is to reduce the latency and increase the accuracy of detections and recognition of objects or events of interest as well as continued proper operation in the event of single or multiple sensor dropouts or mission ground station failures. We propose here an autonomy model that consists of eight levels of autonomy: Remote operation, onboard sensor processing, formation flight, dynamic retasking, onboard sensemaking, dynamic supervised replanning, and dynamic unsupervised replanning. Since space applications have the most stringent constraints on size, weight, and power, it is critical to make decisions that address where the data is stored and how it is processed. We have therefore designed a model that enables specific cognitive functions to be performed on the available data in order to enable the machine decisions that distinguish between the different levels of autonomy.