How to Track Space Objects on Orbit
An ever-increasing number of objects and dynamic observation requirements—complicated by the constant motion of sensors, targets, Earth and the Sun—necessitates adaptive machine learning techniques for tracking space objects.
As more nations and commercial entities develop advanced space capabilities, strengthening Space Domain Awareness (SDA) becomes increasingly critical. U.S. space operators need to be cognizant of everything operating in both Earth’s orbit and beyond. Integrating and streamlining information across the space enterprise is crucial in advancing that effort. Currently, data does not flow smoothly from photons to support tools, delaying real-time decision-making. Aerospace is establishing an operational pipeline for SDA sensors with project Prime Focus — a prototype automated SDA node.
“Current systems rely on human operators, making the pipeline fragile and prone to error,” said project lead Matthew Britton, Principal Engineer in Aerospace’s Space Science Applications Laboratory. “Prime Focus will yield a sustained, daily operational data flow backed by digital engineering and cloud technologies.”
Space control and SDA is a technically challenging problem with many requirements: extensive sensor networks for observational capacity, geometric diversity for solar lighting conditions, broad ground sensor coverage, and response times on a tactical scale.
One foundational issue is combining information from multiple heterogeneous sensors to collect data on a number of resident space objects that is growing exponentially. This requires scheduling sensors, communicating observational tasks to them in real time, acquiring these observations autonomously, and reporting the data back to a central data store. Scheduling is a particularly complex problem with many variables, since the sensors, targets, Earth, and sun are all in relative motion. Artificial intelligence and machine learning techniques offer promising solutions for meeting the necessary timescales.
“An automated, self-scheduling network of sensors cries out for a solution that can accommodate the dynamic requirements imposed by both weather and tactical imperatives,” Britton said.
Prime Focus is an effort to automate the 1-m AeroTel telescope located on top of Aerospace’s laboratory facility in El Segundo, California. AeroTel possesses a range of existing instrumentation developed specifically for SDA, and user inputs generate an automated observation schedule that respects solar lighting conditions.
AeroTel will perform these observations, write data products to cloud storage, and report outcomes to users. The node relies on cloud infrastructure, cloud storage, and software management aligned with modern software practices. The Prime Focus concept can be scaled to fill the SDA need to coordinate large numbers of sensors with multiple targets.
The initial demonstration of Prime Focus was successfully completed in September 2021, illustrating the methodology. Going forward, Prime Focus will serve as a testbed for scheduling, modeling and simulation, artificial intelligence/machine learning and image postprocessing, and will provide a sustained data flow out to external data centers. The goal is to demonstrate an autonomous SDA sensor node that can respond to scheduling events from other research institutions, industry, and government partners. The AeroTel telescope then becomes a discoverable SDA resource to users both in a modern cloud context.
“Prime Focus demonstrates the digital methodology backing a geographically distributed ground sensor network, that scales to large node count,” Britton said. “This type of network addresses our customers’ hard problems in space control and space domain awareness that remain unaddressed by the current SDA architecture.”
