Automated Robust Courses of Action under Non-quantifiable Uncertainty Models (ARCANUM)

ABSTRACT:DECISIVE ANALYTICS Corporation (DAC), in partnership with ANALYTICAL GRAPHICS, Inc. (AGI), proposes to build a novel course of action (COA) development algorithm that allows information-starved space platforms to fulfill their missions when forced to act autonomously. Our effort is entitled Autonomous Robust Courses of Action under Non-quantifiable Uncertainty Models (ARCANUM). Our model is based upon a mathematical notion of robustness that explicitly represents both quantifiable and non-quantifiable uncertainties, thus accounting for sensor data ambiguities and Black Swan events that could not be anticipated or modeled in advance. We will build this approach to be viable within the limited computational resources available onboard a space vehicle, allowing it to reason independently under severe communications restrictions caused by radio frequency interference or adversary electronic/cyber-attacks.BENEFIT:It is no longer viable to design systems that treat space as a sanctuary outside of the confines of the battlespace. To ensure viability in the increasingly non-benign space environment, early threat detection and characterization are crucial so that space vehicles can execute COAs that mitigate these threats. They must choose these actions judiciously, however, balancing mission requirements against resource limitations. If ambiguous sensor readings lead the space vehicle to execute unnecessary maneuvers, it will pre-maturely deplete its fuel supply, compromising its long-term station-keeping capability and shortening its service life. The capability developed under this effort will support the US Air Force (USAF) and the U.S. Strategic Command (USSTRATCOM) in ensuring critical global space missions including Missile Warning, Satellite Communications, and Precision Navigation and Timing.