Sensor Tasking for Spacecraft Custody Maintenance and Anomaly Detection Using Evidential Reasoning

Abstract

Current state-of-the-art sensor tasking for space situational awareness is human-analyst intensive, reactive, and individualized for specific sensors. In application, there are many conflicting objectives and hypotheses that compete for sensor resources, which complicates the sensor assignment problem. Using the Dempster–Shafer theory of evidence, this work proposes a sensor-tasking criterion based on minimizing ignorance in hypothesis resolution. An approach for formulating space situational awareness sensors as evidence-based experts is presented, and an algorithmic implementation for the joint space-object custody and anomaly detection problem is developed and tested. Results show an automated, predictive capability to schedule a sensor network that successfully maintains custody in the presence of poor observation conditions and spacecraft maneuvers. In comparison to covariance minimization, the ignorance-reduction technique resolves hypotheses as well or better while using fewer actions.