SALT LAKE CITY – It’s easy to talk about satellite autonomy but significant work remains to determine exactly which tasks should be handled by machines, according to speakers at the 2025 Small Satellite Conference.
Military aircraft have extensive built-in autonomy thanks to decades of experience identifying useful features in combat exercises. U.S. Space Force satellite operators, in contrast, have little freedom to practice maneuvers because spacecraft have limited fuel reserves and additional satellites won’t be launched for years.
“We need to enable the Space Force to get relieved of that psychological burden and start using their machinery so that they can learn from it, develop [tactics, techniques and procedures] and figure out good thing to automate,” Col. Owen Stephens, U.S. Space Force Space Rapid Capabilities Office (Space RCO) contracting director, said Aug. 12.
During a SmallSat side meeting, panelists agreed that satellite maintenance and routine communications should be automated.
“You’d really like to go lights out for that and only concern people with things people need to be concerned with,” Stephens said.
That’s particularly important given limited government funding and personnel to manage day-to-day satellite operations. Even in a conflict, satellite constellations “need to be operable with a small number of humans,” said Seth Lacy, Air Force Research Laboratory senior scientist for space mobility and precision maneuver. “What can we do to enable our spacecraft to be more functional with fewer humans?”
For instance, autonomy could help satellites evade jamming or other interruptions, said Dean Bellamy, Redwire executive vice president for national security space. “Having that autonomy to continue to provide the data to the soldiers, the sailors, the airmen, the guardians on the ground that need the data is really critical.”
Plus, if machines take over routine tasks, people can figure out “what makes sense to automate and why does it make sense to automate it,” Stephens said.
Automated Collision Avoidance
One task that might make sense to automate is collision avoidance.
Military aircraft have automatic ground collision avoidance systems. But considering whether spacecraft should have autonomous systems to avoid crashing into other satellites or debris raises other questions. For instance, how accurate are the two-line element that tell operators the locations of space objects?
“Are they accurate enough to enable the autonomous nature of a system like that?” Stephens asked. “The answer right now is probably no.”
Massive Compute
Government organizations are investing in technologies needed for autonomy.
“We’re trying to use machine learning to come up with optimal maneuvering strategies with regard to a noncooperative vehicle,” said Benjamin Bahney, Lawrence Livermore National Laboratory space program leader. “There needs to be a technological push from people that know how to run algorithms on massive compute to get answers that help satellite operators understand what’s physically possible.”
Communications capacity is another potential bottleneck. As space-based sensors become more capable and spacecraft operate further from Earth, operators will have trouble sending all the data they capture to the ground for processing.
Lawrence Livermore is developing wide-field optical telescopes to survey cislunar space and process imagery onboard satellites to reduce demand for bandwidth.
In addition, space-based edge computing capabilities must continue to improve.
“That is a technology problem that still is not solved,” Bahney said.