Space-Based Sensor Tasking Using Deep Reinforcement Learning

Abstract

To maintain a robust catalog of resident space objects (RSOs), space situational awareness (SSA) mission operators depend on ground- and space-based sensors to repeatedly detect, characterize, and track objects in orbit. Although some space sensors are capable of monitoring large swaths of the sky with wide fields of view (FOVs), others—such as maneuverable optical telescopes, narrow-band imaging radars, or satellite laser-ranging systems—are restricted to relatively narrow FOVs and must slew at a finite rate from object to object during observation. Since there are many objects that a narrow FOV sensor could choose to observe within its field of regard (FOR), it must schedule its pointing direction and duration using some algorithm. This combinatorial optimization problem is known as the sensor-tasking problem. In this paper, we developed a deep reinforcement learning agent to task a space-based narrow-FOV sensor in low Earth orbit (LEO) using the proximal policy optimization algorithm. The sensor’s performance—both as a singular sensor acting alone, but also as a complement to a network of taskable, narrow-FOV ground-based sensors—is compared to the greedy scheduler across several figures of merit, including the cumulative number of RSOs observed and the mean trace of the covariance matrix of all of the observable objects in the scenario. The results of several simulations are presented and discussed. Additionally, the results from an LEO SSA sensor in different orbits are evaluated and discussed, as well as various combinations of space-based sensors.