You are here

Evaluating Data Strategies in Training AI Solutions for Space C2




The technology within this topic is restricted under the International Traffic in Arms Regulation (ITAR), 22 CFR Parts 120-130, which controls the export and import of defense-related material and services, including export of sensitive technical data, or the Export Administration Regulation (EAR), 15 CFR Parts 730-774, which controls dual use items. Offerors must disclose any proposed use of foreign nationals (FNs), their country(ies) of origin, the type of visa or work permit possessed, and the statement of work (SOW) tasks intended for accomplishment by the FN(s) in accordance with the Announcement. Offerors are advised foreign nationals proposed to perform on this topic may be restricted due to the technical data under US Export Control Laws.


OBJECTIVE: Evaluate and illustrate the consequences of employing various data strategies in training autonomous systems (Artificial Intelligence and Deep Learning based algorithms).  Demonstrate how space superiority may depend on employing AI which is sufficiently trained and how space domain awareness data can be used to support this training.


DESCRIPTION: Autonomous space command and control systems are already being fielded within mega-constellations and are being considered for other space systems in deep space.  Today flight dynamics teams are informed by third party space situational awareness products which are derived from different approaches to collecting data on operational spacecraft and debris.  As increased activity which is considered to be dual-use becomes more prevalent, we will increasingly see the use of autonomy in addressing various challenges for which human cognition may not scale to or be capable of responding to in time.  A key step in the development of AI solutions is the training of the algorithms which will do the decision-making.  This effort seeks to evaluate the relative performance of such algorithms when they are trained by space domain awareness data of varying quality, density, geometric diversity, precision, and timeliness.


The past few years have seen successful proximity operations in GEO, a rapid increase in maneuverable traffic in LEO, and more technology options for autonomous systems to perform enhanced services in space including life extension, refueling, inspection, etc.  As more interactions between spacecraft are observed, there is an increasingly comprehensive body of data which can be used to train algorithms which enable autonomous space command and control (C2) solutions.  We are already seeing evidence of this training possibly being employed today.  As digital twins of space systems become available, it will be increasingly possible to model space systems, their onboard logic, and to enable training within a digital environment.  As new systems are fielded, their autonomous logic can be further trained on-orbit and it will be desirable to compare on-orbit experience to models used to train within the digital environment.  This topic seeks to develop the digital training environment for space C2 algorithms, and to demonstrate how the agents within this environment can be trained using real-world and simulated SDA data.


PHASE I: Design a digital traning environment which can enable modeling and simulation using digital representations of space systems, as well as data-driven reconstructions of real-world operations informed by SDA data.


PHASE II: Develop a prototype digital training environment which can enable modeling and simulation using digital representations of space systems as well as data driven reconstructions of real-world operations informed by SDA data.  Using this environment, train AI algorithms for space command and control and demonstrate the relative performance of these agents as the data strategies used for training are varied.  Evaluate the impact of training autonomous systems using SDA data of varying quality, density, geometric diversity, precision, and timeliness.


PHASE III DUAL USE APPLICATIONS: Integrate prototype solution into systems available in operational environments for operator and analyst evaluation and feedback.  Expected TRL at Phase III entry is 5.



  1. Kyriakopoulos, George & Pazartzis, Photini & Koskina, Anthi & Bourcha, Crystalie. (2021). ARTIFICIAL INTELLIGENCE AND SPACE SITUATIONAL AWARENESS: DATA PROCESSING AND SHARING IN DEBRIS- CROWDED AREAS.;


KEYWORDS: ai/ml; artificial intelligence; space domain awareness; space c2

US Flag An Official Website of the United States Government