Predictive sensor tasking and decision support for space situational awareness using evidential reasoning

Space situational awareness;Evidential reasoning;Decision support;Sensor tasking

Jaunzemis, Andris Davis ; Holzinger, Marcus J. Aerospace Engineering Feigh, Karen Johnson, Eric Blake, Travis Luu, Kim ; Holzinger, Marcus J.

University:Georgia Institute of Technology

Department:Aerospace Engineering

关键词: Space situational awareness;    Evidential reasoning;    Decision support;    Sensor tasking;   

Others  :  https://smartech.gatech.edu/bitstream/1853/59921/1/JAUNZEMIS-DISSERTATION-2018.pdf

美国|英语

来源: SMARTech Repository

【 摘 要 】

Situation awareness is the perception of elements in the environment, comprehension of their meaning, and projection of their status into the future. Space situational awareness (SSA) is particularly concerned with accurately representing state knowledge of space objects to resolve potential threats, such as collision. Tracking techniques used in the space surveillance system still rely largely on models and applications from the 1950s and 1960s, while the number of tracked objects continues to grow with improved sensor technologies and ease-of-access to space. This work frames the SSA sensor tasking problem to interrogate specific hypotheses using evidential reasoning. First, cognitive systems engineering practices are applied to derive cognitive work and information relationship requirements for SSA decision-support systems and provide insight on the utility of hypothesis-based methods in SSA. To evaluate hypothesis-based methods for SSA, the spacecraft anomaly detection problem is formulated as a binary hypothesis test using distance metrics while accounting for non-Gaussian boundary conditions to improve applicability to non-linear orbital dynamics. Next, a sensor tasking criterion is developed to gather the evidence that minimizes ambiguity, or ignorance, in hypothesis resolution. The application of evidential reasoning provides a rigorous framework for quantifying ambiguity and allows inclusion of diverse SSA sensors. Building upon this method, a generalized evidence-gathering framework, Judicial Evidential Reasoning (JER), is proposed for hypothesis resolution tasks. JER also accounts for confirmation bias by applying a principle of equal effort. Resource allocation is a non-linear, high-dimensional, mixed-integer problem, so JER also applies adversarial optimization techniques to address computational tractability concerns. Finally, a prototype SSA decision support system is developed based on the derived requirements to evaluate workload and situation awareness impacts of hypothesis-based tasking. This work aims to enable predictive sensor tasking to provide decision-quality information and improve decision-maker situation awareness and workload.