Which Observation Comes Next? Starling's Satellites Make a Shared Plan

When one satellite finds an interesting signal, what should its partners do? Repeating the observation may help, but watching something else may contribute more to the shared objective. The problem is not merely getting every spacecraft to collect data. Separate observations have to inform a common plan while conditions change. NASA's Distributed Spacecraft Autonomy experiment, or DSA, tested this decision-making layer aboard the Starling mission. [1][2]

Starling consisted of four CubeSats carrying several technologies for spacecraft cooperation. DSA concerned distributing scientific observation work and reconsidering plans when new information arrived. This account examines NASA reports from May 2024 and February 2025, with a knowledge cutoff of 13 October 2026. Future objectives in those dated reports are not automatically treated as subsequently completed flight results. [1][2]

A changing subject for observation

The experiment investigated Earth's ionosphere. The spacecraft received GPS signals passing through that region, providing information about features worth observing. This was not simply a story about using GPS to locate a satellite on a map. The environment through which the signals traveled was itself the scientific subject. NASA explained that some features were fleeting, making rapid information exchange valuable. [1]

The spacecraft also continually changed position relative to one another, the GPS satellites, and the ionospheric environment. One vehicle's best recent observation might not be sufficient to determine all the group's subsequent work. When new information arrived, its value and the opportunities available to other vehicles had to be reconsidered. That differs from carrying out a long, fixed list of commands in sequence. [1]

In DSA, spacecraft evaluated their own results, exchanged information, and reached a shared approach to observations. New information prompted analysis of new observation and action plans. The useful capability was responding to changing conditions without requiring a ground team to select every detail after every measurement. Both NASA reports emphasized this distribution of scientific work. [1][2]

Supplying an objective is not supplying every step

Autonomy did not mean removing humans from the mission altogether. People supplied the scientific objective, software, and operating framework. What distinguished the experiment was that ground operators did not preprogram every individual science observation. The spacecraft could determine their observational activities together within the objective provided. A programmed capacity to make decisions is not an unprogrammed system. [1]

This also limits the metaphor of shared brains. The satellites were not becoming a biological mind, acquiring human intentions, or inventing their scientific purpose. Distributed reasoning and planning functions were being tested. A spacecraft processing local information did not have to act in isolation; collaboration connected local assessment with a common goal.

NASA described the demonstration period as August 2023 through May 2024. That interval should not be mistaken for a launch date or a statement that every experiment began simultaneously. Likewise, the four-spacecraft mission description does not establish that every member performed the same role throughout every experiment. Keeping mission scale separate from detailed participation avoids inventing an operating arrangement absent from the account. [1]

A network and a science plan have different jobs

Starling's Mobile Ad-hoc Network experiment, MANET, tested communication links and the transfer of commands and data. The May 2024 report described information being routed through another swarm member to a spacecraft experiencing difficulty with direct ground communications. That was a concrete benefit of cooperative networking. Finding a path for a message, however, was not the same task as deciding which scientific observation deserved priority. [2]

DSA therefore did not replace communications, and communications alone did not produce the scientific plan. A group whose members can all exchange messages might still duplicate an observation without useful purpose. What information should be shared, and how work should change in response, require an additional decision process. The experiment's significance lay in managing those relationships rather than simply increasing the number of satellites.

Other experiments on the mission addressed complementary functions. StarFOX used optical observations for orbital information, while ROMEO investigated onboard maneuver planning. The May account stated that ROMEO had demonstrated determining and planning an orbital change, with execution of maneuvers still being developed. Reading the mission's success headline as proof that every possible autonomous maneuver had already been performed would conflict with that more specific account. [2]

Where did the sixty lunar spacecraft fly?

A second branch of DSA work investigated scaling to a larger group near the Moon. That experiment was conducted on the ground. Virtual small spacecraft and rack-mounted flight computers operated in simulated lunar orbits. The aim was to assess cooperation for position, navigation, and timing services as the number of vehicles increased. [1]

NASA's February 2025 account reported close to one hundred tests over two years and feasibility for a simulated swarm of up to sixty spacecraft. This did not mean sixty satellites had been delivered to lunar orbit. Using flight computers can make a test more concrete than a purely conceptual exercise, but does not automatically reproduce a completed orbital mission with actual environmental and communications conditions. [1]

The Earth-orbit demonstration and virtual lunar scalability study consequently supplied complementary evidence without becoming interchangeable. One concerned shared observational decisions in flight. The other explored a larger arrangement in simulation. A future objective of working with hundreds of spacecraft remained a further development step, not another result already established by those tests. [1]

A bounded result about shared decisions

Distributed operation offers a way to reduce dependence on an individual vehicle. The reports nevertheless do not establish immunity to every failure, flawless operation at an unlimited scale, or the disappearance of human oversight. They describe successful experiments rather than certifying the reliability of all future swarms. No numerical improvement in scientific return can be inferred simply from the use of several spacecraft. [1][2]

The demonstrated direction is more specific and more useful. Observations were not merely collected for transmission to Earth; they could inform the choice of subsequent shared work. The link between local measurement, exchanged information, and revised plans was tested in flight. Starling points toward something different from sending the same instruction to more satellites: preserving a common objective while the group reorganizes its activities in response to conditions.

Sources

[1] Figliozzi, G. NASA (4 February 2025). NASA Demonstrates Software ‘Brains’ Shared Across Satellite Swarms. https://www.nasa.gov/centers-and-facilities/ames/nasa-demonstrates-software-brains-shared-across-satellite-swarms/

[2] Friesen, T. NASA (29 May 2024). Swarming for Success: Starling Completes Primary Mission. https://www.nasa.gov/directorates/stmd/swarming-for-success-starling-completes-primary-mission/