SBIR-STTR Award

As space becomes increasingly congested and contested space operators must rapidly assess threats with confidence to know what actions can be taken. Machine learning (ML) offers promise in efficiently dealing with these highly complex systems, however a major challenge is producing ML systems which are both robust and adaptable. Applied Defense Solutions (ADS) and the University of Texas at Austin propose to develop a Robust Adaptive Machine Learning (RAM) architecture for supporting decision making processes in the context of space battle management command and control. ADS will utilize ML techniques, both supervised and unsupervised, such as automated structure learning, physics guided data science, and active learning with user feedback. ADS will develop an architecture with predictive capability, based on past observation of patterns of life, with the ability to build relational correlations between disparate sources of data. ADS proposes five demonstration use cases to apply this RAM architecture. Training RAM algorithms requires large quantities of high quality data and ADS has access to unique space situational awareness data from its Global Optical Network.